1

From what I understand,

Core web vital assessment (the top section of PageSpeed Insight result) uses field data and the lighthouse section uses lab data. And the core web vital assessment is what influences my search rankings.

Often metrics from the web vital are better than the lighthouse, for example, an LCP of 2.2s vs 4.2s. And lighthouse results are very inconsistent between Chrome with different locations/machines and one from PageSpeed Insight with the same throttle settings.

Which metrics are telling the truth? Core web vital or the lighthouse in PageInsight? Can I assume that the result from the lighthouse on my machine is different from the lighthouse in PageInsight due to the environment regardless of the same throttle settings? On my machine, most of the pages are 90+ when PageInsight's lighthouse shows less than 60.

Thanks!

2 Answers 2

4

There's a great post about exactly this titled Why lab and field data can be different (and what to do about it).

Which metrics are telling the truth? Core web vital or the lighthouse in PageInsight?

This is the distinction between lab and field data.

When you run a Lighthouse test, you need to configure it to run at a predetermined connection speed. The default speed for mobile tests in PSI is different from desktop tests. The default speed for a mobile test in PSI may be different from the one you use to test locally.

However, with field data like the CWV assessment in PSI powered by the Chrome UX Report, it's an aggregation of many (if not all) real-user experiences on whatever their connection speeds happen to be.

So if by "truth" you're looking for the true speed experienced by real users, then you should rely on field data.

Then what's the purpose of lab data? To help you identify opportunities to improve your field data.

When the lab test results show that LCP is 4.2 seconds, it's telling you that all of its recommendations are based on a simulated user experience with an LCP of that speed. If your real users tend to experience LCP at 2.2 seconds instead (at the 75th percentile), know that you're optimizing for some of the worst experiences on your site. It's a good thing to try to make the slowest experiences faster, but don't be fooled into thinking that an LCP of 4.2 seconds from the lab is typical of real users' experiences.

1

There is no one truth because it depends on many factors, such as web connection speed, latency from the connecting user, origin / destination country, number of hops, hardware (firewalls, switches, CPU speed), software (OS, browser, extensions, antivirus, etc.), external scripts, etc.

From my personal experience, I use several tools at once:

From there, I can get a better picture to analyze where I need to focus my optimization efforts.

1
  • I think this question is more about whether to trust lab vs field results, rather than which lab tools give the most realistic results. Lab realism is entirely dependent on the performance characteristics of the particular page being tested and the configuration of the test, so there's no easy one-size-fits-all lab tool that provides the "truth". That's why field data is considered the source of truth. May 5 at 14:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.