In an SEO office-hours video Google’s John Mueller answers why the PageSpeed Insights score and the Search Console Core Web Vitals scores are consistently different.
He explains why two measurements of page speed never match and why it makes sense that they don’t match.
PageSpeed Insights vs. Search Console CWV Report
It’s reasonable to ask why two things that purport to measure the same thing provide different results.
This is the question:
“When I check my PageSpeed Insight score on my website, I see a simple number. Why doesn’t this match what I see in Search Console in the Core Web Vitals report?
Which one of these numbers is correct?”
Different Page Speed Scores
John Mueller first mentions that the scores provided by PageSpeed Insights and Search Console are actually different sets of numbers.
“I think maybe first of all, to get the obvious answer out the door, where is no correct number when it comes to speed, when it comes to understanding how your website is performing for your users.
In PageSpeed Insights, by default, I believe we show a single number that kind of is a score from 0 to 100, something like that, which is based on a number of assumptions, where we assume that different things are a little bit faster or slower for users.
And based on that we calculate a score.
In Search Console we have the Core Web Vitals information which is based on three numbers for speed, responsiveness and interactivity.
And these numbers are slightly different, of course, because it’s three numbers not just one number.”
The Method for Calculations is Different
Mueller next explains how the scores for the two measurements (PageSpeed Insights and Search Console CWV report) are derived and why that results in different scores.
“But also there’s a big difference in the way that these numbers are determined.
Namely there’s a difference between so-called field data and the lab data.
Field data is what users have actually seen when they go to your website. And this is what we use in Search Console.
That’s what we use for search as well.
Whereas lab data is kind of a theoretical view of your website, like where our systems have certain assumptions where they think, well the average user is probably like this, using this kind of device and with this kind of a connection perhaps.
And based on those assumptions, we will estimate what those numbers might be for an average user.
And obviously you can imagine those estimations will never be a hundred percent correct.”
So what Mueller is saying then is that Search Console scores reflect what actual site visitors saw.
The way Google measures those actual numbers is through visitors who have opted in to allowing Chrome to provide anonymized core web vitals data.
Google doesn’t measure every site visitor, only those who have opted in to sending that data to Google.
Contrasting with Search Console’s field data, PageSpeed Insights is creating a simulation of what a user might experience.
The purpose of Search Console data is to show what real site visitors are experiencing.
The purpose of PageSpeed Insights data is to provide an estimate of what’s going on in order to provide diagnostic feedback on what may be causing poor webpage speed performance.
Mueller next comments on how even though Search Console is based on real site visitors that the data is going to vary and not be consistent.
“And similarly, the data that users have seen, that will change over time as well, where some users might have a really fast connection or a fast device and everything goes really fast on their website, or when they visit your website.
And others might not have that.
And because of that, this variation can always result in different numbers.
Our recommendation is generally to use the field data, the data you would see in Search Console, as a way of understanding what is kind of the current situation for a website.
And then to use the lab data, namely the individual tests that you can run directly yourself, to optimize your website and try to improve things.
And when you are pretty happy with the lab data that you’re getting with your new version of your website, then over time you can collect the field data, which happens automatically and double-check that users actually see it as being faster or more responsive as well.
So in short, again, there is no absolutely correct number when it comes to any of these metrics.
There is no kind of like absolutely correct answer where you’d say, this is what it should actually be.
But rather, there’s different assumptions and different ways of collecting data and each of these are subtly different.”
Two Metrics for Two Purposes
Perhaps a helpful way to think about the two metrics is that they both are useful for two different purposes.
Search Console Core Web Vitals data is derived from actual visitors and provides a sampling of what the user experience of real-world site visitors is (field data0. Actual site visitor data might highlight issues that may not show up in a simulated sampling.
PageSpeed Insights scores are simulated scores whose purpose is for analysis and debugging page speed issues. This tool identifies problems that interfere with fast page downloads and offers suggestions on how to fix them.
The data sources are different so they will never match exactly. But that doesn’t matter because the intention for each tool is different.
Google’s official documentation for the Search Console CWV report states:
“The Core Web Vitals report shows URL performance grouped by status, metric type, and URL group (groups of similar web pages).”
The purpose of PageSpeed Insights tool is stated on the webpage of the tool:
“Make your web pages fast on all devices”
In short, this is what each tool is for:
- The purpose of the Search Console tool is to provide granular snapshots of real-world site performance.
- The purpose of PageSpeed Insights is diagnostic, to identify problems and offer suggestions for improvement.
Watch John Mueller answer the question at the 31 second mark.