At a glance
- PageSpeed Insights shows field data for "the previous 28-day collection period" at the 75th percentile, next to one Lighthouse lab run.
- CrUX includes opted-in Chrome users on desktop and Android. Chrome on iOS, Android WebView and other Chromium browsers are excluded.
- Pages need to be publicly discoverable and popular enough to appear in CrUX; noindex pages are not eligible.
- Lighthouse's standard mobile test simulates 150 ms latency, 1.6 Mbps download and a 4x CPU slowdown.
Two kinds of data
Google's PageSpeed Insights documentation puts it simply: lab data "is useful for debugging issues, as it is collected in a controlled environment", while field data is "useful for capturing true, real-world user experience". The lab data comes from "a simulated load of a page on a single device" with fixed network conditions; the field data comes from many visits on many devices and networks.
| Field (CrUX, your RUM) | Lab (Lighthouse, DevTools) | |
|---|---|---|
| Source | Real visits | One scripted load |
| Period | Rolling 28 days in PageSpeed Insights | The moment of the test |
| INP | Measured | Not measured on load; Total Blocking Time is a proxy |
| Logged-in and personalized states | Included as visitors see them | Only if you script them |
| Best use | Deciding what to fix and confirming a fix | Finding the cause and testing a change |
Who CrUX includes
Chrome's methodology page lists the conditions: users must have usage statistics reporting enabled, sync their browsing history without a sync passphrase and use a supported platform, which means desktop Chrome on Windows, macOS, ChromeOS and Linux, and Chrome on Android. Chrome on iOS is not included, nor are Android apps using WebView or other Chromium browsers such as Edge.
For a casino operator this matters. If many of your mobile players use iPhones, CrUX shows you the Android part of your mobile audience only. Your own real-user monitoring can fill the gap, with the caveat that some metrics are not available in every browser.
Pages must also be publicly discoverable and "sufficiently popular". A page that returns noindex, through a meta tag or an X-Robots-Tag header, is not eligible. Logged-in areas such as the cashier, account and game pages behind login rarely appear in CrUX at URL level. When a page lacks data, PageSpeed Insights falls back to origin-level data for the whole site.
Why the two disagree
- Devices and networks. The lab simulates one mid-tier phone; your players use a spread of devices.
- Cache state. Lighthouse loads with an empty cache. Returning players often have the CSS, fonts and scripts cached.
- Personalization and consent. Field visits include logged-in lobbies, market variants, consent choices and tags that fire after consent. A lab run sees the default variant.
- Interactions. INP comes only from real interactions, so the lab cannot report it from a page load.
- Timing. Field data lags a release by up to 28 days. A lab improvement today will not show fully in field data for weeks.
Reading a Lighthouse score
The Lighthouse performance score is a weighted mix of five lab metrics. For Lighthouse 10, Chrome's documentation lists Total Blocking Time at 30%, Largest Contentful Paint at 25%, Cumulative Layout Shift at 25%, and First Contentful Paint and Speed Index at 10% each. The same page notes that "a lot of the variability" in scores between runs "is not due to Lighthouse". Run it several times and compare medians, and never report one run as a site's performance.
A sound workflow
- Start with field data by template and device type (Search Console groups similar URLs).
- Pick a representative URL for the failing group and profile it in the lab on a throttled mobile setting.
- Fix one cause, repeat the lab test under the same settings and record the result as a lab result.
- Wait for field data to confirm. Our measurement protocol sets out what to record.
Questions
Why does PageSpeed Insights show no field data for my page?
The page may not have enough eligible visits, or may be noindex. PageSpeed Insights then falls back to origin data or shows none.
Can a staging site have field data?
Not in CrUX. Staging sites are usually noindex and private, which makes them ineligible. Use lab tests and your own monitoring there.
Is Total Blocking Time the same as INP?
No. web.dev calls TBT a reasonable lab proxy for INP but "not a substitute" for it.
Sources
- Google for Developers, About PageSpeed Insights, last updated October 21, 2024.
- Chrome for Developers, CrUX methodology, last updated June 20, 2024.
- Chrome for Developers, How Lighthouse calculates your overall Performance score, footer date September 19, 2019 (lists Lighthouse 10 weights).
- GoogleChrome, Lighthouse throttling documentation, no date shown.
- web.dev, Total Blocking Time (TBT), last updated October 15, 2025 (footer shows November 4, 2024).
- Google Search Console Help, Core Web Vitals report, no date shown.