Hi all,
We just had a support question regarding how to evaluate and use 4SEO real Core Web Vitals (CWV) measurement, and after replying to it, I thought this might be a good topic to talk about, again. The question was similar to:
4SEO reports a bad Core Web Vitals score for many pages but common online speed test tools give me a perfect score. What's going on here?
There's still a bit of confusion on how Core Web Vitals are used by Google, how important they are (hint: not that much) and how to use the numbers provided.
What are Core Web Vitals?
A quick reminder first: Core Web Vitals is Google's method for evaluating websites performance. These measurements are used when Google ranks pages in search results, which is why they get a lot of attention.
Google has used speed to rank pages for a long time, but in 2021 they started using Core Web Vitals, a more realistic method, better at evaluating what real users experience when visiting a website.
Last year I made a short videos explaining the inner working of Core Web Vitals, so I won't dive too deep into the details of CWV here. Please refer to that video to learn quickly what CLS, LCP or FID are (yes, web people love acronyms).
In short, Core Web Vitals attempt to represent the actual experience of visitors to a given page of your site with just 3 numbers :
- LCP: how long before the page loads?
- FID: how much time before I can click on a button and it does something?
- CLS: is the page content moving around when images or videos finish loading?
From Google standpoint, a website either Passes or Fails the CWV test. You pass the test if your CWV numbers are above given values. If you pass, you get a "bonus" over sites that don't.
The major difference with previous methods, Google's or other performance test sites, is that CWV are measured by the visitor browser, in real time, when they visit your page.
GoogleBot, the Google robots that crawls your site, does not measure Core Web Vitals.
Are Core Web Vitals an important SEO ranking factor
To put it bluntly: Not that much.
Google has stated repeatedly that performance is more like a tie-breaker - or maybe just a bit more important than that but not much.
If 2 pages are considered both an excellent fit to reply to somebody searching something in Google, then if one passes CWV and the other doesn't, the "fast" page will likely perform better in search results.
So it does matter, especially if you are in a very competitive field. But it cannot be stressed enough that it's your content (and links, internal and external, and structured data and ...) that will be the major part of the ranking decision, not speed. Also, of course, it's only an advantage if your competitors are slow!
But know that CVW are a pass/fail kind of thing: the "pass" value for LCP for instance is 2.5 seconds. If your LCP score is 4 seconds, you fail. If your LCP score is 2.5 seconds, you pass. And if your LCP score is 0.8 seconds - which is very, very fast, excellent, congratulation! - then you pass too, but no more than the site with a 2.5 seconds LCP. There's no SEO benefit in having the fastest faster website of all websites: 0.8 seconds is the same as 2.5 seconds.
That said, having a fast site is good for your visitors, and it's one of these cases where your's and Google's wishes are in perfect alignment I think.
So what was the question again?
4SEO reports out of bound, failing CLS. But if I go to pagespeed.web.dev or gtmetrix.com, these sites report perfectly CSL at zero, or very low. Why is that and what should I do?
Which tool should I rely on? And what tool does Google use to determine my CLS, or is that unknown?
A perfectly valid question, if you asked me - well, they asked me actually :) - and so here is my response, a bit redacted for clarity:
4SEO reports out of bound, failing CLS. But if I go to pagespeed.web.dev or gtmetrix.com, these sites report perfectly CSL at zero, or very low. Why is that and what should I do?
The thing to understand here is that the number when YOU measure CLS is not important. CLS - as used by Google for SEO ranking - is the CLS measured when your visitors view your website.
Which tool should I rely on? And what tool does Google use to determine my CLS, or is that unknown?
It is very well-known and they don't use a tool. They use Chrome. So if you use Chrome and you view a page at Weeblr.com, then Chrome will measure the performance you saw when visiting that page and send that back to Google.
And so if you try to load a Weeblr.com page while driving through a tunnel, with extremely slow mobile network and maybe on an old phone, Google will record a very poor CWV score because the page took 10 seconds to load. Please don't do that by the way, you will be lowering our CVW score!!!
But if the next day you load the same page on your top of the line phone, while connected to your super fast Wifi, the exact same page will get a very high CWV score.
The same page, 2 different scores, depending on what device you are using and the network connection you were going through.
Some important things to know about Core Web Vitals
- CWV is not one measurement: measurements are taken over many visitors, and aggregated over the last 28 days. The scores taken into account are updated daily, using data measured over the previous 28 days). So one single measurement on a test site is not a good representation of what Google will use.
- Google - and 4SEO - use the measurements 75th percentile, which means if your CLS is 0.42, 75% of your site visitors saw a CLS of 0.42 or better. Again, one single measurement on a test site does not represent that properly
- 4SEO uses the exact same code Google is using in Chrome to measure Core Web Vitals (the library is open source)
- Contrary to Google, we show you all the data we have, even if there was only one measurement taken, while Google will wait to have a large (but unknown) number of data points to start taking into account measured CWV as a ranking factor
- It is unclear what they do when they don't have actual CWV data from real users. They said they can extrapolate from other pages, but how that's done is not known.
- You can see the CWV values they use in your Google Search Console. That's the only source of truth for CWV in SEO.
- All other tools are debugging tools, trying to identify why your perf is good or not, but they are not actually measuring CWV as used by Google. They can't because they are not embedded in your website. Use these tools to debug CWV issues. You can start with just your browser developer tools.
- 4SEO is the only way you can have something close to reality for CWV, because it's embedded in your website and measures actual users viewing pages.
Special note for 4SEO users: if you have only a few measurements for a page, the displayed number can be biased. More measurements should even out the values: if you only have one visitor on a page, and they were loading that page on your site on an old phone, under a tunnel, while driving (don't browse and drive though), 4SEO will display that single poor CWV score. Google instead will likely discard that value, or not use it until they have a 1000 reading or something.
Closing note
That particular user then checked their Google Search Console Core Web Vitals scores and here is what they found:
My speed is good (all green check marks, also in Google Search Console), but the CLS is too high at about 650 out of 1700 pages (according to GSC, and about 400 according to 4SEO)
So these pages, getting a perfect CLS score in GTMetrix, PageSpeed, etc were indeed considered as failing (on desktop) by Google - and 4SEO.
Hope that was useful, please feel free to ask questions in the comment!
Cheers,
Yannick