You open PageSpeed Insights and your site scores 62 on mobile. The same week, Search Console says your URLs are rated "Good". Or the other way round: Lighthouse gives you a 95 and Search Console flags pages in red. The natural reaction is to assume one tool is wrong, but neither is lying: they measure different things, in different ways, at different times.
Understanding that difference helps you avoid two expensive mistakes: chasing a score that doesn't reflect what your visitors actually experience, and treating a site as fine when in practice it's slow for plenty of people.
What each one measures: simulation versus real users
Lighthouse (the engine behind PageSpeed Insights and the Lighthouse tab in Chrome DevTools) loads your page once, in a controlled environment, with a simulated device and network. That's lab data: a score from 0 to 100 and a list of specific issues, reproducible but limited to a single load.
Field data comes from the Chrome User Experience Report (CrUX): statistics from real Chrome users who have opted in to sharing them, on their actual phones and networks. It feeds the Core Web Vitals report in Search Console and the real-user experience block at the top of PageSpeed Insights. For each metric the 75th percentile is used, meaning the value that three out of four visits meet or beat.
Why the two numbers shouldn't match
- Device and network: the lab uses a fixed profile; the field mixes powerful phones, older devices, wifi and 4G.
- Cache: the lab starts from scratch; many real visitors already have resources stored.
- Interaction: the lab doesn't click or type, so it only approximates responsiveness with indirect metrics. The field does record it.
- Time: the lab is a snapshot of today; the field is an average of recent weeks.
That's why you can have a 95 in the lab and problems in the field (your visitors use slower phones than you assume) or a 62 in the lab and good field results (your audience is mostly desktop on a good connection).
Which one to use for what: debugging versus measuring
Use the lab to debug: it tells you which image is too heavy, which script blocks rendering or which element jumps around, and lets you check straight away whether your fix works. Use the field to measure for real: it shows whether your visitors have a good experience, and it's the reference Google uses to assess page experience.
A sensible workflow: check Search Console to see which groups of URLs have real problems, run Lighthouse on a representative page from that group to find the cause, fix it, and wait for the field data to confirm.
The 28-day window and low-traffic sites
CrUX data is calculated over a rolling 28-day window. Even if you fix something today, the field report won't fully reflect it for roughly four weeks, because visits from before the change keep counting. It's not a bug: it's how long the average takes to refresh.
On top of that, CrUX only publishes data when there are enough visits from Chrome users. If your site gets little traffic, Search Console may say there isn't enough data, or PageSpeed Insights may show only origin-wide data rather than data for the specific URL. Then the lab is all you have to go on: treat it as guidance, not a verdict.
Setting a realistic target without obsessing over 100
The Lighthouse score is a weighted summary of several metrics and can vary from one run to the next without you touching anything. Chasing 100 usually costs a lot of effort for an improvement nobody notices. It's more useful to set goals around what matters:
- Field metrics in the "Good" range on your key pages (home, services, product pages).
- No serious, obvious lab issues: unoptimised images, heavy third-party scripts, layout shifts.
- A trend that improves after each change, always comparing the same device type and the same pages.
Frequently asked questions
Why does PageSpeed Insights give me a different score every time?
Because it simulates a single load, and factors such as network or server load at that moment introduce small variations. Run the test several times and look at the trend, not a single number.
Does Google penalise me for a low Lighthouse score?
The score itself isn't what's assessed. What matters for page experience is the field data for the Core Web Vitals metrics. Lighthouse is for diagnosing, not an official measure.
My site has little traffic and Search Console shows no data. What should I do?
Use Lighthouse to fix the obvious problems and check back from time to time to see whether field data has appeared. With low traffic it's normal for the report to take a while to gather enough information, or never to have it for some URLs.