Partly. Google measures a short list of things real Chrome users encounter, reads the rest of the page as a document, and infers the remainder from its own logs. The line between those three is the useful part.
Is there a user experience score?
No. There is no single number Google assigns to a site's user experience. Google's own documentation on page experience describes a set of aspects its ranking systems look at, and states that there is no one combined page experience signal. That matters practically: there is no dial to move, so "improve page experience" is not a task. The individual aspects are tasks, and they are measured in different ways with different lag.
What parts of experience does Google actually measure?
| Aspect | How Google gets it | What it still cannot see |
|---|---|---|
| Loading, layout stability, interaction responsiveness | Core Web Vitals from Chrome field data | Other browsers, and any page with too few visits to report |
| Mobile rendering | Googlebot rendering the page itself | Whether the layout is pleasant to use once it renders |
| HTTPS | The certificate on the request | Nothing subjective at all |
| Intrusive interstitials | Analysis of the rendered page | A popup that fires only for signed-in or returning visitors |
Everything in the left column is mechanical. Everything a designer would call experience sits in the right column.
Where does the field data come from?
The Chrome User Experience Report, which aggregates measurements from Chrome users who opted in to sharing usage statistics, over a rolling 28-day window. Three consequences follow. The report lags, so a fix shipped today moves it over weeks rather than overnight. It reflects the devices and networks of your actual visitors, not your laptop on office wifi. And a page with too few visits never gets its own data, so Search Console reports it inside a group of similar URLs whose numbers may be dominated by pages you were not thinking about. Note also that Interaction to Next Paint replaced First Input Delay as a Core Web Vital in March 2024, so measurement advice written before then is aiming at a metric that no longer exists.
Google measures what a browser can report and what its own result pages record. Everything between the click and the conversion is invisible to it, which means that work is worth doing for revenue, never for rankings.
Do bounce rate and time on page feed rankings?
No, and the reason is structural rather than a matter of policy. Bounce rate and session duration are properties of your analytics tool. They live in your own property, they are computed by rules you configured, and nothing crawls them. Google has stated repeatedly that it does not use Google Analytics data for ranking, and there is no mechanism by which it could. The correlation people notice is real but backwards: a page that matches intent earns both better engagement and better rankings from the same cause.
Does Google use click behavior at all?
Yes, in its own logs, which is a different claim from watching your site. Documents and testimony released during the United States antitrust case against Google described ranking systems that draw on aggregated click and query data from Search itself. That is Google's record of its own result pages, not a feed from your pages. The practical consequence is narrow and useful: your title and description sit inside that loop, because they are what earns or loses the click in organic search. Your scroll depth does not.
Is E-E-A-T something Google measures?
No. E-E-A-T is a concept from the search quality rater guidelines, the document Google gives the human evaluators who score sample results. Those scores are used to judge whether a change to the ranking systems made results better. They do not adjust the ranking of the page that was rated. So there is no E-E-A-T value to raise. What exists are the on-page facts raters are told to look for: who wrote this, what can be shown about what they know, whether the claims are checkable, and who stands behind the site.
What does Google miss entirely?
Anything behind a login, anything after the conversion step, form friction, checkout flow, support quality, and the state of a page on a browser Chrome does not report. Comment threads and reviews are user-generated content that can change a page's real experience daily while Google only re-reads it on recrawl. Where experience is ambiguous, Google falls back on what it can parse without a human: headings, links and structured data. That is not Google being lazy. It is the limit of what a crawler can know.
So what should you fund?
Split the backlog in two. Items that move a field metric or a rendered-page check have a mechanical path to rankings and belong to search work with a measurable before and after. Items that improve the experience after the click have no such path and should be justified on conversion, retention or support load, where they usually win on their own merits anyway.
The mistake is not doing either kind of work. It is putting the second kind on a search roadmap, promising rankings for it, and then explaining the flat graph six months later. Decide now which column each open ticket belongs in, and who gets to judge whether it worked.