CrUX vs PageSpeed Insights — the real difference

CrUX (Chrome User Experience Report) and PageSpeed Insights (PSI) often get conflated, but they're not the same thing. PSI is a tool that reads from CrUX (plus runs a Lighthouse test). CrUX is the underlying dataset. Knowing which is which saves you from chasing the wrong number.

The short answer

CrUX = the dataset of real-world Chrome performance measurements, aggregated by Google. It's what Google uses for ranking signals.

PageSpeed Insights = a free Google web app that:

Reads CrUX field data and displays it
Also runs a fresh Lighthouse audit in a controlled lab environment
Combines both and gives you a single Performance score (0–100)

So PSI shows you CrUX data alongside lab data. The CrUX numbers in PSI are the same numbers in our tool. The lab score is something extra.

What CrUX measures

Field data from real Chrome users worldwide, on the actual sites they visit, over a rolling 28-day window. Per origin (or for high-traffic URLs, per URL). Anonymized and aggregated — Google never tells you which user got which result.

LCP, INP, CLS, FCP, TTFB
Form-factor split (phone, tablet, desktop)
28-day p75 values
Updated monthly (typically the second Tuesday)

CrUX is what Google's ranking algorithm actually uses for the page-experience signal. It's the source of truth.

What PSI's lab score measures

A one-shot Lighthouse run from a Google server. Standardized environment: mid-tier mobile CPU, 4G network throttle, no browser extensions, no real user state. Reproducible — you can run it 100 times and (mostly) get the same number.

The lab score is great for debugging. You can change one thing, re-run, and see if it helped. But it's not what users experience and it's not what Google ranks on.

Why CrUX and lab numbers disagree

They almost always disagree. Common reasons:

Real users have caches, lab runs don't. Lab runs hit your origin cold every time. Real users have your CSS, fonts, and images warm in their cache.
Real users are on a distribution of devices. Lab runs are on Google's standardized "Moto G Power" simulation. Your actual users are everywhere from 3-year-old Androids to M3 iPads.
Real users see real third-party scripts. Lab runs see your scripts in isolation; real users have your scripts plus their ad blocker, their password manager, their tab-suspender extension.
CrUX is 28-day p75. Lab is a single run. One Lab number might be 1.2s while CrUX p75 is 3.8s — both are correct measurements of different things.
Lab can't measure INP. Lab tests INP as TBT (Total Blocking Time) which is a proxy, not the same metric. CrUX measures actual interactions.

When to use which

Use CrUX (or our tool) when:

You want to know "is this passing Core Web Vitals for SEO?"
You're benchmarking against competitors
You're tracking a trend over weeks/months
You want the number Google actually sees

Use PSI's Lighthouse lab score when:

You're debugging a specific page
You're testing a code change before/after
You don't yet have enough Chrome traffic for CrUX data
You want detailed diagnostics (waterfall, opportunities, etc.)

The best workflow: check CrUX for the field reality, then use Lighthouse to debug specific pages or measure changes you're about to ship.

What PSI's "Performance score" actually is

The 0–100 score in PSI is a weighted average of lab metrics: FCP, LCP, TBT, CLS, Speed Index. The exact weights have changed over the years. As of 2025: LCP 25%, TBT 30%, CLS 25%, FCP 10%, Speed Index 10%.

That score is not what Google uses for rankings. The ranking signal is the CrUX p75 values vs Google's thresholds. A site can score 65 on PSI lab but still pass Core Web Vitals if real users' p75 is good.

Why our tool just shows CrUX

We deliberately don't show a lab score on this site. Two reasons:

The CrUX number is what matters for rankings. Adding a lab score creates noise about which one is "right."
Lab runs are expensive. They take 15–30 seconds and they're not deterministic across server load. Our tool returns CrUX data in <1s from a cache that's already there.

If you want the lab score, PageSpeed Insights is the canonical tool — we link to it from every domain page.

What to do with the disagreement

If your lab score is great but your CrUX is poor: your code is fast in isolation, but real users are seeing something else. Usually third-party scripts, slow networks, or device CPU variance. Audit your script tags, profile on a real low-end Android, and look at INP specifically.

If your CrUX is great but your lab score is poor: real users are getting cached content. Your fresh load is slow but most users never see it. Worth fixing for new visitors, but the SEO signal is already in your favor.

Either way: trust CrUX as the ground truth for rankings. Use the lab score as a debugging aid, not a target.

By Paulo de Vries · Published 2026-05-20