#and like one day i may expand the covers tab to include that stuff but it's a lot bigger undertaking that will take a lot more research
Explore tagged Tumblr posts
lesbiancarat · 10 months ago
Note
hello!!!! i just stumbled upon your carat resource carrd and want to thank you for putting in the effort and time for such a well made and informative library of seventeen content!! i noticed that there are a few songs missing: seungkwan's cover of 'text me merry christmas' w/ AKMU suhyun, and the HHU songs from the diamond edge tour (joker - wonwoo solo, what's the problem, and madness maxed out)
also i think the miss you by sicboy ft. vernon MV is missing from the carrd :)
ahh thank you! i try my best to get everything but sometimes little things slip through the cracks so i appreciate you letting me know! i can't believe i missed some of these lol. these should all be added now!
Music -> Covers -> Text Me Merry Christmas
Music -> Unofficial & Unreleased Songs -> Joker / What's the Problem / Madness Maxed Out
Music -> Collaborations -> Miss You
9 notes · View notes
Text
HP ELITE DRAGONFLY (2020)
Tumblr media
A year ago's HP Elite Dragonfly was maybe the most pleasant business PC we've at any point tried. The current year's Elite Dragonfly is practically indistinguishable from that machine, which is fine in light of the fact that, once more, the HP Elite Dragonfly was remarkable in pretty much every manner.
This new arrangement to a great extent feels and looks equivalent to a year ago's model; it's as yet one of the sleekest, chicest business workstations available, and it has the most attractive plan of any convertible that HP right now sells.
HP has changed four things. To begin with, the Dragonfly is currently 5G-empowered, however that component isn't coming until mid-2020. Second, it has a coordinated Tile tracker, which is coming to models in mid-May. Third, it has another protection situated screen that incorporates HP's most recent Sure View Reflect innovation. Fourth, its mechanical parts are presently for the most part worked from reused materials.
The new Dragonfly is slender, light, delightful, and pretty much impeccable. You can get the base arrangement for around $1,500, yet the model we're taking a gander at costs $2,179, which is a serious sticker price. (This particular model doesn't appear to be accessible on HP's site yet, yet arrangements with comparative specs, including Tile, an i7, vPro, and the Sure View Reflect screen are in the $2,100 to $2,700 territory, contingent upon RAM, stockpiling, and different highlights.) The new highlights work, however they're extravagances, not necessities, for by far most of individuals. In case you're a C-Suite power client who's consistently in a hurry, they may be a commendable spending cost for you. In any case, you'll likely be okay with a less expensive EliteBook except if cash is actually no item for you or your organization.
Tumblr media
HP has been staging in reused materials into the Elite Dragonfly ridiculous year. It's the first ultrabook to fuse sea bound plastics (that is, plastic litter gathered in sea territories, which would somehow have wound up in the sea). The organization said a year ago's model would consolidate 50% PCR plastics (and 5 percent sea bound plastics) in its speaker box and 35 percent PCR plastics in its bezels.
The organization's objectives, from that point forward, have gotten more yearning. It reported at CES 2020 that more than 80% of the Dragonfly's mechanical parts and 90 percent of the magnesium suspension are currently made of reused materials. This activity isn't explicit to Dragonfly; HP says that other new HP Elite and HP Pro PCs will fuse the new composite segments.
The new material hasn't ruined the undercarriage in any capacity. I can't recollect the last time I held a PC this light that felt this durable. (HP didn't react to different requests about the specific weight and measurements of this unit, however it's around 2.5 pounds.)
There's no flex in the console and practically none in the showcase. The suspension additionally feels extremely pleasant to the touch; the magnesium is smooth, and the adjusted edges and corners mean you never get jabbed. Fingerprints are frequently a concern on dull items, however the wrist rests and console remained sans print following a few days of utilization. The touchpad and top amassed a few, however I could just see them under splendid light.
HP has likewise traded out the plastic covers on the console (the material is presently 50% sourced from reused DVDs) and the screen's bezels (presently 35% reused plastic). The keycaps are a piece plasticy yet at the same time feel extraordinary, and the bezels don't appear to be any unique from those on the old model.
The new Dragonfly closely resembles an exceptionally decent PC. Also, hello, presently it's more practical.
Tumblr media
The second enormous change is the new screen, which utilizes HP's Sure View Reflect to expand client security. Reflect is the fourth era of SureView, which the organization dispatched with its EliteBook 1050 G3 in 2016. You can in any case get the Dragonfly arranged with more seasoned boards, including the low-power 1W presentation we tried a year ago, a showcase with SureView Gen3, and a HDR 400 showcase with 3840 x 2160 goal. This is only an extra choice.
TheThe Dragonfly has two screen modes, which you can helpfully flip between by squeezing F2. There's Sharing Mode, which conveys up to 773 nits of brilliance and wide review points, and Privacy Mode, which colors the screen with the end goal that individuals passing by can't sneak around.
With Privacy Mode on at 50% splendor, I was unable to see a thing on the screen while sitting opposite to it. I started to make out certain substance at the edge of the showcase at a 45-degree point, yet I didn't get a decent look until I was nearly dealing with it directly. In case you're utilizing Privacy Mode in broad daylight, dubious figures will experience difficulty spying except if they're discernibly peering near. Somebody sitting close to you on a transport or train may make out bits, however they will not draw near to the full picture.
Note, however, that the screen gets dimmer in Privacy Mode. It additionally kicks back significantly more glare, particularly underneath most extreme brilliance. Indeed, even with the screen at max, the glare was generous enough that the gadget wasn't usable outside — and even inside, I wouldn't have needed to utilize it under 50% brilliance. It's so natural to flip back to Sharing Mode, however, that I'm not very animated about that compromise.
Fingerprints do stay on this board (it's a touchscreen) and were difficult when I attempted to clear them off. It was somewhat irritating, however such force clients considering this setup may like to utilize HP's Active Pen, which ships with this unit.
With regards to extravagant screen includes, the inquiry is consistently whether they'll affect battery life. To my alleviation, not exclusively is the new Dragonfly's battery life amazing, however it's quite better than that of the late-2019 model. I took the ultrabook (on the default Better Battery power profile and 50 percent brilliance) through my typical workday of shuffling eight to 12 Chrome tabs, running Slack, web based Spotify and YouTube, and a periodic Zoom call, parting time about similarly between Sharing Mode and Privacy mode. I got 11 hours and 38 minutes, which is the best battery life result I've at any point gotten from a PC. In the event that you need an item that can dependably chip away at the go, you'll experience difficulty discovering better compared to the Dragonfly.
The Dragonfly's really pivotal (yet less business-arranged) new component is the Tile mix. Tile, for those new, makes little Bluetooth-empowered gadgets that you can join to your keys, tote, wallet, or different assets. On the off chance that you lose the Tile-associated object, you can find it utilizing the Tile application on your telephone. This is the main PC with a Tile tracker worked in, which can help you discover the Dragonfly if it's lost or taken.
What's shrewd about the Tile is that it's fueled by its own equipment separate from that of the Dragonfly. That implies it can sound an alert through its own coordinated speaker in any event, when the PC is off. The tracker draws a modest quantity of force from the PC, however, HP didn't react to our requests about the points of interest of the relationship. In the event that the Dragonfly is off when you lose it, HP says the Tile will continue to work for 20 days. On the off chance that the PC's in hibernation, you actually have 2.5 days.
Setting up the Tile is an extremely straightforward interaction. I needed to actuate the gadget through the Tile Microsoft application (it comes preloaded onto the Dragonfly) and make a record with the help. (The Tile plays a great jingle while you're setting it up.) Once I downloaded the Tile application on my telephone and signed in, I was set.
At the point when you open the Tile versatile application, you'll see a rundown of any Tile items you've associated, including the Dragonfly. In the event that you select the PC and press "Discover," its tracker will sound a boisterous caution. (iPhone clients can likewise do this with a Siri alternate way.) You can likewise see its keep going known area on a guide. In the event that you lose the journal outside of Bluetooth range, you can assign it as "lost" and initiate Tile's Community Find highlight, which will send you a caution with its area at whatever point another Tile passes inside its Bluetooth range. On the off chance that you register for Tile Premium, you can get more highlights, including a more extended 30-day area history, and Smart Alerts which advise you in the event that you've gone out without your PC (or another gadget joined to a Tile).
At the point when I concealed the Dragonfly in a heap of clothing in a wardrobe, I could hear the tracker's alert quite well when I was in a similar room, and it was perceptible (however I needed to listen hard) from the following room over. Outside, it was perceptible until around 60 feet away. The Tile would in general remain associated with my telephone until I was around 140 feet away. Those are similar outcomes to those you can anticipate from the independent Tile Pro tracker.
A coordinated Tile tracker is not really a fundamental component for the normal business client. Be that as it may, on the off chance that you travel a ton and you need the innovation, it works.
We can speedrun through the remainder of the standard PC stuff since it's equivalent to the Dragonfly Elite that we investigated in late 2019. The console is a flat-out homer, with superb travel and almost no clamor. The glass trackpad is comparably great, with a smooth surface and a peaceful snap. The port determination is amazing for a particularly lightweight 2-in-1, including two USB-C Thunderbolt 3 ports, an HDMI port, and a 3.5mm sound jack on the right, just as a USB-A port, a Kensington lock port, and opening for a SIM card on the left. Windows Hello fills in as it ought to, and the webcam has a helpful protection shade. The four-speaker exhibit conveys probably the best encompass sound I've at any point heard from a Windows PC. The PC stays cool significantly under substantial burdens, and the fans aren't noisy in any way.
The new Dragonfly additionally has a similar eighth Gen vPro-empowered Core i7-8665U processor as a year ago's, model. (tenth Gen Comet Lake vPro isn't out yet; we should see that in the not-so-distant future.) It's a bit of a disappointment not to see tenth Gen contributes a particularly expensive machine — the tantamount Comet Lake tenth Gen i7 has six centers, so you're missing out on multithreaded execution ability, which is valuable for assignments like incorporating code and doing elaborate things with Excel. eighth Gen chips are additionally in a difficult situation contrasted with Ice Lake chips, which will improve inventive work, on account of Intel's new Iris Plus Graphics. The 8665U actually took care of my ordinary burden (which for the most part incorporates Chrome tabs and Slack) fine and dandy, yet in the event that your workday incorporates additional requesting assignments, you might be in an ideal situation hanging tight for a Dragonfly model with tenth Gen CPUs.
Other downsides of the past model continue. The force button holds its inconvenient situation on the left side; I inadvertently squeezed two or multiple times when I was hefting the PC around. The screen actually has the 16:9 viewpoint proportion, which is confined for profitability use; I regularly needed to zoom out to 80 or 70 percent to serenely utilize tabs next to each other. A PC focusing on business power clients definitely should be 16:10 or, shockingly better, 3:2.
These are little blemishes contrasted with the things the Dragonfly does extraordinarily well, which is fundamental to all the other things. In case you're on the lookout for a vPro-empowered framework with both an underlying Tile GPS beacon and an incorporated security screen, this is the solitary PC available with that mix of highlights. In the event that you and your organization are super-rich, sure, spend lavishly away.
Be that as it may, this Dragonfly is an extravagance item. It's the Galaxy S20 Ultra or the iPhone 11 Pro Max of business workstations; it's amazing, yet a great many people needn't bother with it. In the event that you can abandon the extravagant highlights, a less expensive arrangement will turn out great. Furthermore, in the event that you needn't bother with vPro and will go for a ThinkPad all things being equal, you can get an advanced processor.
1 note · View note
nuggetisawesome · 4 years ago
Text
Protect Your Browsing/Data Sharing
For free, because in this house I don’t believe in making people pay for basic human rights.
I’ve decided to share my browsing protecting tips here. Digital security is usually quite expensive, but it doesn’t have to be! In this day and age, you’ll be told to watch out for home-born hackers and ‘hacktivists’ accessing your data, but I gotta tell you, what your own governments and ISPs are doing makes this more important (aka: they’re worse). 
I know with all the TIKTOK IS SPYING ON YOU stuff, a lot of my friends have come to me seeking some advice on this. This is also great if you don’t want parents checking your browsing >_> just sayin’
If anyone has questions - drop me an ask! I’ll always answer for this topic. I am also happy to ‘expand’ on one of these suggestions if they’re unclear :) 
Note: This works under the assumption you have your default ISP provided router and can’t get another one for whatever reason. I will advise that if you can get an additional router, do so! Try to avoid the one the ISP has provided to you. 
Additional Note: This is not ‘optimal’. There is no such thing in security �� everything has a backdoor. 
Let’s get cracking! This is a long, and thorough post, but I _do _have a pdf somewhere if you want it because it looks nicer :*) 
Use Tor to browse. 
There you go, there’s my advice leaves
https://support.torproject.org/ to Download/Install/Run.
Don’t change anything, except maybe using Tor in ‘bridge’ mode.
Okay, you can use other browsers (see: Chrome/Firefox), but they are not as secure as Tor.
USE A VPN IF YOU ARE GOING TO USE TOR! I prefer Firefox (extensions + good security)
________________________________________________________________
Install the following extensions if you have Chrome or Firefox:
Privacy Possum
Stops tracking cookies. PSA: Cookies are not evil, certain cookies can be.
Firefox:  https://addons.mozilla.org/nl/firefox/addon/privacy-possum/
Chrome: https://chrome.google.com/webstore/detail/privacy-possum/ommfjecdpepadiafbnidoiggfpbnkfbj
Ghostery
Stops tracking adverts and cookies. Why do I need this in addition to Privacy Possum? Ghostery specifically looks at tracking cookie ads. It’s like adding MOAR POWAH to Privacy Possum.
Firefox: https://addons.mozilla.org/nl/firefox/addon/ghostery/
Chrome: https://chrome.google.com/webstore/detail/ghostery-%E2%80%93-privacy-ad-blo/mlomiejdfkolichcflejclcbmpeaniij?hl=nl
HTTPS Everywhere
Enforces HTTPS. If you look next to the URL in your browser, you’ll see the little lock which indicates the specific URL is secure and uses HTTPS. Many websites still use the old HTTP, which is not as secure and you should NEVER EVER VISIT AN HTTP SITE ITS LIKE READING A BOOK OVER SOMEONE’S SHOULDER, thank you.
Firefox: https://addons.mozilla.org/nl/firefox/addon/https-everywhere/
Chrome: https://chrome.google.com/webstore/detail/https-everywhere/gcbommkclmclpchllfjekcdonpmejbdp?hl=nl
Adblock Plus
Foff, ads.* Firefox: https://addons.mozilla.org/nl/firefox/addon/adblock-plus/
Chrome: https://chrome.google.com/webstore/detail/adblock-plus-free-ad-bloc/cfhdojbkjhnklbpkdaibdccddilifddb
________________________________________________________________
DNS Settings
Ideally, you should change this on your router. ISPs use a default DNS – you don’t want to use anything those bastards say you should use. 
Use Cloudflare or OpenDNS:
Cloudflare is more secure overall and keeps up to standards in addition to not storing your data, whereas OpenDNS is great at avoiding malicious websites, just take your pick really 😊There are a ton of other options, feel free to google ‘free DNS servers’. Google has it’s own as well, but, yaknow, it’s Google.
Cloudflare
Primary Server: 1.1.1.1
Secondary Server: 1.0.0.1
OpenDNS
Primary Server: 208.67.222.222
Secondary Server: 208.67.220.220
Add these to your router settings:
In a browser, go to http://192.168.1.1/ or http://192.168.1.0/ (it varies per router). This will lead to your router’s configuration portal. Don’t have a router with a configuration portal? Throw it in the trash and tell your ISP they suck for giving it to you.
Login to the admin portal. If you have not configured this or set a password, try the default combinations: usernames are usually ‘admin’ or blank, the passwords are usually blank, ‘admin’, or ‘1234’.
Each router is different, navigate to where it asks for DNS values or servers, and enter the above addresses. You will see ‘Static’ near the DNS options, select it. This also ensures you’re in the right place. If you’re not sure what to do, look up the model/make of your router and check how you can change DNS.
Whilst you’re at it, change your WiFi password from the default one, and create a proper password for the WiFi portal login. If these two things are kept as default, all these protection methods are pointless as it is easy to crack your router passwords.
Can’t do this on your router because your parents are ds?* No worries! This can be done on your device! :) Yeah, I know how parents work. 
Windows OS
Go to Control Panel <Network and Internet < Network and Sharing Center
Click on the link next to “Connections:”* Click “Properties” in the dialogue that pops up.
Select Internet Protocol Version 4 < Click Properties < Select “Use Following DNS Servers” < Enter the primary and secondary server addresses
Do this again for Internet Protocol Version 6 in the list.
Boom. Windows is so nice to make this easy.
Mac OS
Go to Apple Menu < System Preferences < Network
Select the Network you’re connected to
Click Advanced
Select DNS Tab
Click the + button < Enter chosen DNS < Save
Linux OS
I’m going to assume if you’re using Linux, you know how to use the terminal and are using a modern Linux system. Enter these line by line. There are many ways to do this (Google is your friend)
·        sudo apt update
·        sudo apt install resolvconf
·        sudo systemctl status resolvconf.service (check that it is running)
·        sudo systemctl start resolvconf.service (to start it, use ‘enable’ instead of ‘start’ to enable)
·        sudo nano /etc/resolvconf/resolv.conf.d/head
·        nameserver YOUR.DNS.ADDRESS.HERE
·        nameserver YOUR.SECOND.DNS.ADDRESS.HERE
·        sudo systemctl start resolvconf.service
Android
Oh yeah, you can do this on phones too wiggles eyebrows. Note, if you’re using a VPN it will lock you out of editing this. Turn it off, edit your DNS, turn it back on. This can be tricky with mobile devices that have not been jailbroken (I don’t advise doing that if you have no clue what you’re doing).
Go to Settings < Connections < WiFi
Select the gear icon next to your current WiFi
Select Advanced < Ip Settings drop-down < Static
Enter chosen DNS options under “DNS 1” and “DNS 2”
iPhone
Go to Settings < Wi-Fi
Select the arrow button next to your current WiFi
Select DHCP tab, scroll down to DNS
Select DNS, and enter your DNS servers
TEST YOUR DNS IS WORKING:
OpenDNS: https://welcome.opendns.com/ (You’ll see a “Welcome to OpenDNS” message”
Cloudflare: https://www.cloudflare.com/ssl/encrypted-sni/ (You’ll see check marks for all fields)
________________________________________________________________
Turn of WPS on router. Enable encryption on router.
If you can access your router portal, find any sort of toggle or field that says “WPS” and disable it. WPS= bad.  
Wherever there is an option for WPA2 (or higher) to enable, enable it.
Enable the firewall on your router and Operating System – ALWAYS. If you disable this, you’re disabling an additional layer of security. Firewalls are confusing things and a royal pain in the ass to configure, but having the default is better than having nothing.
________________________________________________________________
Configure your browsers. 
Browsers have most things enabled by default, including tracking your location, turning your microphone on etc. Let’s disable that nonsense and make them ask you for permission because it’s 2020.
I’m using Chrome as an example below because it is INFAMOUS for this. Essentially, go through your browser and scroll through settings you don’t like.
Go to the little menu icon < select “Settings”
Sign out if it’s linked to your Google account. Let’s not give Chrome a reason to track your browsing history for your account >_>* Disable EVERYTHING:
Tumblr media
Here, I turned off everything I would find annoying except autocomplete because I’m lazy.
Tumblr media
NICE TRY GOOGLE, YOU CANT SAVE MY CREDIT CARD. (Seriously, don’t ever EVER autosave passwords/payment info).
Tumblr media Tumblr media Tumblr media
The trick is to balance ease of use with security. These may vary from person to person, in general though, if there’s a setting ‘ask before etc.’ select that over ‘allow’. 
As a rule of thumb:
NEVER ENABLE FLASH (not even an  ‘ask before’), NEVER ENABLE LOCATION (ask before is fine, but at your own risk), NEVER ENABLE CAMERA (ask before is fine, but at your own risk, use the desktop version of an application over the browser version), NEVER ENABLE MICROPHONE (same as camera)
________________________________________________________________
Additional Tipss
Check every single social media setting. You should disable anything that accesses your privacy, if possible.
Cover your camera with a sticker. Disable it unless it’s needed
Disable your mic unless you need it.
Turn off Bluetooth/NFC when you don’t need it.
Have two separate networks/Wi-Fi for smart devices and personal devices.
Always use two/multi-factor-authentication for every single app, site, device etc. that you can.
Biometrics are preferable and the usual chosen default (fingerprints, retina scan, NOT FACE OR VOICE THIS IS SO EASY TO IMMITATE BRO PICTURES ARE A THING)
SMS (try to avoid if you can, please)
App ‘token’ authentication. A good choice if done well.
Hard physical key or token. The best option (Google: Yubikey, for some information on how this works).
Passwords
I know you use the same password for everything – get a centralized password manager, and start using different ones. Examples include PasswordSafe, Keeper, Bitdefender etc. Try go for a Cloud provider, and pay a little bit for the extra security and backup. If they’re compromised, then you will know, and you’ll be able to change everything. You can tie password managers to a token too.
USE PASSPHRASES, 17 characters is a good average length, use a mix of characters, uppercase, lowercase, numbers, ascii etc.
It doesn’t matter if your password is ‘complex’, it matters if it is complex and long. Servers and computers these days are jacked up on tech steroids and can bruteforce many things, given enough time.
Anti-virus.
EVERYONE SHOULD HAVE ONE, ON EVERY DEVICE. If you have a device that can add AV, add it. This goes for phones, PCs, smartTVs, you name it. 
Free versions are okay, some free ones I like are Bitdefender, Kaspersky, McAfee, Avast (hate their fihsfirstg89ewjg9srjgrd ads though).
Sorry Mac users, that belief that you don’t need one is from 2008. Windows has more security built in than Mac, which means Mac devices should 100% make sure they are adding an AV. 
VPN
Ahhhhh. The great VPN. A tricky one. Most free versions I find incredibly slow, but give them a try – play around! A VPN is an excellent addition and these days, I’d argue it’s an absolute must. Many AV solutions include a VPN package with their deal. If you want to make sure those sites don’t share your data, this is the thing that will hurt the most - a good VPN will make it a jumbled mess. 
Updates - just do them.
There is no complete security in this day and age – it really is just a matter of time. If you use social media, you’re traceable, be it by the company, ISP, some bored 10 year old, or your ex, your data is out there circulating. Once it’s on the internet, it’s there forever, so don’t worry too much and try to make sure it’s all complex binary trash so that they open it and go “WTF” 😊
11 notes · View notes
forlawfirmsonlymarketing · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast: We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°). How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords. This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4: Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate. The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency: Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story: If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone. Now, let’s pick three different data points (all of these are from the top 20): From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes. There’s an even weirder story buried in the May 2020 data. Consider this: LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update): Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%. Now let’s look at Google Play, which appeared to be a clear winner after two days: You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update. How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means: While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history. Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions. Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers: Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change: Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains. Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after: It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions. Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time. Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).   Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
https://www.businesscreatorplus.com/googles-may-2020-core-update-winners-winnerers-winlosers-and-why-its-all-probably-crap/
0 notes
isearchgoood · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
via Blogger https://ift.tt/2WsXYgc #blogger #bloggingtips #bloggerlife #bloggersgetsocial #ontheblog #writersofinstagram #writingprompt #instapoetry #writerscommunity #writersofig #writersblock #writerlife #writtenword #instawriters #spilledink #wordgasm #creativewriting #poetsofinstagram #blackoutpoetry #poetsofig
0 notes
evempierson · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
0 notes
gamebazu · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
https://ift.tt/2AjddzJ
0 notes
thanhtuandoan89 · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
0 notes
daynamartinez22 · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
0 notes
camerasieunhovn · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
0 notes
ductrungnguyen87 · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
0 notes
nutrifami · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
0 notes
paulineberry · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
0 notes
forlawfirmsonlymarketing · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast: We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°). How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords. This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4: Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate. The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency: Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story: If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone. Now, let’s pick three different data points (all of these are from the top 20): From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes. There’s an even weirder story buried in the May 2020 data. Consider this: LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update): Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%. Now let’s look at Google Play, which appeared to be a clear winner after two days: You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update. How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means: While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history. Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions. Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers: Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change: Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains. Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after: It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions. Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time. Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).   Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
https://www.businesscreatorplus.com/googles-may-2020-core-update-winners-winnerers-winlosers-and-why-its-all-probably-crap/
0 notes
kjt-lawyers · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
0 notes
epackingvietnam · 5 years ago
Text
Google's May 2020 Core Update: Winners, Winnerers, Winlosers, and Why It's All Probably Crap
Posted by Dr-Pete
On May 4, Google announced that they were rolling out a new Core Update. By May 7, it appeared that the dust had mostly settled. Here’s an 11-day view from MozCast:
We measured relatively high volatility from May 4-6, with a peak of 112.6° on May 5. Note that the 30-day average temperature prior to May 4 was historically very high (89.3°).
How does this compare to previous Core Updates? With the caveat that recent temperatures have been well above historical averages, the May 2020 Core Update was our second-hottest Core Update so far, coming in just below the August 2018 “Medic” update.
Who “won” the May Core Update?
It’s common to report winners and losers after a major update (and I’ve done it myself), but for a while now I’ve been concerned that these analyses only capture a small window of time. Whenever we compare two fixed points in time, we’re ignoring the natural volatility of search rankings and the inherent differences between keywords.
This time around, I’d like to take a hard look at the pitfalls. I’m going to focus on winners. The table below shows the 1-day winners (May 5) by total rankings in the 10,000-keyword MozCast tracking set. I’ve only included subdomains with at least 25 rankings on May 4:
Putting aside the usual statistical suspects (small sample sizes for some keywords, the unique pros and cons of our data set, etc.), what’s the problem with this analysis? Sure, there are different ways to report the “% Gain” (such as absolute change vs. relative percentage), but I’ve reported the absolute numbers honestly and the relative change is accurate.
The problem is that, in rushing to run the numbers after one day, we’ve ignored the reality that most core updates are multi-day (a trend that seemed to continue for the May Core Update, as evidenced by our initial graph). We’ve also failed to account for domains whose rankings might be historically volatile (but more on that in a bit). What if we compare the 1-day and 2-day data?
Which story do we tell?
The table below adds in the 2-day relative percentage gained. I’ve kept the same 25 subdomains and will continue to sort them by the 1-day percentage gained, for consistency:
Even just comparing the first two days of the roll-out, we can see that the story is shifting considerably. The problem is: Which story do we tell? Often, we’re not even looking at lists, but anecdotes based on our own clients or cherry-picking data. Consider this story:
If this was our only view of the data, we would probably conclude that the update intensified over the two days, with day two rewarding sites even more. We could even start to craft a story about how demand for apps was growing, or certain news sites were being rewarded. These stories might have a grain of truth, but the fact is that we have no idea from this data alone.
Now, let’s pick three different data points (all of these are from the top 20):
From this limited view, we could conclude that Google decided that the Core Update went wrong and reversed it on day two. We could even conclude that certain news sites were being penalized for some reason. This tells a wildly different story than the first set of anecdotes.
There’s an even weirder story buried in the May 2020 data. Consider this:
LinkedIn showed a minor bump (one we’d generally ignore) on day one and then lost 100% of its rankings on day two. Wow, that May Core Update really packs a punch! It turns out that LinkedIn may have accidentally de-indexed their site — they recovered the next day, and it appears this massive change had nothing to do with the Core Update. The simple truth is that these numbers tell us very little about why a site gained or lost rankings.
How do we define “normal”?
Let’s take a deeper look at the MarketWatch data. Marketwatch gained 19% in the 1-day stats, but lost 2% in the 2-day numbers. The problem here is that we don’t know from these numbers what MarketWatch’s normal SERP flux looks like. Here’s a graph of seven days before and after May 4 (the start of the Core Update):
Looking at even a small bit of historical data, we can see that MarketWatch, like most news sites, experiences significant volatility. The “gains” on May 5 are only because of losses on May 4. It turns out that the 7-day mean after May 4 (45.7) is only a slight increase over the 7-day mean before May 4 (44.3), with MarketWatch measuring a modest relative gain of +3.2%.
Now let’s look at Google Play, which appeared to be a clear winner after two days:
You don’t even need to do the math to spot the difference here. Comparing the 7-day mean before May 4 (232.9) to the 7-day mean after (448.7), Google Play experienced a dramatic +93% relative change after the May Core Update.
How does this 7-day before/after comparison work with the LinkedIn incident? Here’s a graph of the before/after with dotted lines added for the two means:
While this approach certainly helps offset the single-day anomaly, we’re still showing a before/after change of -16%, which isn’t really in line with reality. You can see that six of the seven days after the May Core Update were above the 7-day average. Note that LinkedIn also has relatively low volatility over the short-range history.
Why am I rotten-cherry-picking an extreme example where my new metric falls short? I want it to be perfectly clear that no one metric can ever tell the whole story. Even if we accounted for the variance and did statistical testing, we’re still missing a lot of information. A clear before/after difference doesn’t tell us what actually happened, only that there was a change correlated with the timing of the Core Update. That’s useful information, but it still begs further investigation before we jump to sweeping conclusions.
Overall, though, the approach is certainly better than single-day slices. Using the 7-day before-vs-after mean comparison accounts for both historical data and a full seven days after the update. What if we expanded this comparison of 7-day periods to the larger data set? Here’s our original “winners” list with the new numbers:
Obviously, this is a lot to digest in one table, but we can start to see where the before-and-after metric (the relative difference between 7-day means) shows a different picture, in some cases, than either the 1-day or 2-day view. Let’s go ahead and re-build the top 20 based on the before-and-after percentage change:
Some of the big players are the same, but we’ve also got some newcomers — including sites that looked like they lost visibility on day one, but have stacked up 2-day and 7-day gains.
Let’s take a quick look at Parents.com, our original big winner (winnerer? winnerest?). Day one showed a massive +100% gain (doubling visibility), but day-two numbers were more modest, and before-and-after gains came in at just under half the day-one gain. Here are the seven days before and after:
It’s easy to see here that the day-one jump was a short-term anomaly, based in part on a dip on May 4. Comparing the 7-day averages seems to get much closer to the truth. This is a warning not just to algo trackers like myself, but to SEOs who might see that +100% and rush to tell their boss or client. Don’t let good news turn into a promise that you can’t keep.
Why do we keep doing this?
If it seems like I’m calling out the industry, note that I’m squarely in my own crosshairs here. There’s tremendous pressure to publish analyses early, not just because it equates to traffic and links (frankly, it does), but because site owners and SEOs genuinely want answers. As I wrote recently, I think there’s tremendous danger in overinterpreting short-term losses and fixing the wrong things. However, I think there’s also real danger in overstating short-term wins and having the expectation that those gains are permanent. That can lead to equally risky decisions.
Is it all crap? No, I don’t think so, but I think it’s very easy to step off the sidewalk and into the muck after a storm, and at the very least we need to wait for the ground to dry. That’s not easy in a world of Twitter and 24-hour news cycles, but it’s essential to get a multi-day view, especially since so many large algorithm updates roll out over extended periods of time.
Which numbers should we believe? In a sense, all of them, or at least all of the ones we can adequately verify. No single metric is ever going to paint the entire picture, and before you rush off to celebrate being on a winners list, it’s important to take that next step and really understand the historical trends and the context of any victory.
Who wants some free data?
Given the scope of the analysis, I didn’t cover the May 2020 Core Update losers in this post or go past the Top 20, but you can download the raw data here. If you’d like to edit it, please make a copy first. Winners and losers are on separate tabs, and this covers all domains with at least 25 rankings in our MozCast 10K data set on May 4 (just over 400 domains).
Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!
#túi_giấy_epacking_việt_nam #túi_giấy_epacking #in_túi_giấy_giá_rẻ #in_túi_giấy #epackingvietnam #tuigiayepacking
0 notes