PageSpeed 100 Doesn’t Mean Your Site Is Fast – Here’s What Does
Written by Sorin Gavenea, founder of Gavenea Studio
Published: 29/06/2026 · Updated: 09/07/2026
A PageSpeed Insights score of 100 doesn’t mean your website is fast. It means your site performed well in a single lab test on one simulated device. The number Google actually ranks you on is a different one entirely: field data – the real experience of your visitors over the last 28 days. The two routinely disagree, which is why a site can show a perfect PageSpeed score and still fail Core Web Vitals where it counts. Understanding the gap between the two is the difference between optimising for a vanity number and optimising for the speed your buyers and Google actually see.
The two numbers PageSpeed Insights is actually showing you
Most people open PageSpeed Insights, see a big green 100, and close the tab satisfied. But the report shows you two completely separate things, and the one everyone fixates on is the less important of the two.
(Passed / Failed), at the top
What it is:
Field data – real users
Where it comes from:
Chrome User Experience Report (CrUX), last 28 days
Performance score, below it
What it is:
Lab data – one synthetic test
Where it comes from:
Lighthouse, a single simulated run
The colourful 0–100 number is the lab score. The quiet “Passed” or “Failed” line above it is the field verdict – and that’s the one tied to how you rank. It’s entirely possible to score 98 in the lab and read “Failed” in the field assessment right above it. Most people never notice, because the big number pulls the eye.
What the lab score actually measures
The Lighthouse score is a controlled estimate. It runs your page once, on a simulated mid-tier phone, over a throttled network connection, in a clean environment with nothing else going on. That makes it repeatable and genuinely useful for diagnosis – you can change one thing, re-run it, and see the effect immediately.
But it is not your audience. It’s one machine, one run, one set of conditions. And it has a specific blind spot: responsiveness. The lab measures Total Blocking Time (TBT) as a stand-in, but TBT only estimates how busy the main thread is during load – it can’t replicate the dozens of real taps, clicks, and scrolls that determine your real Interaction to Next Paint (INP). A page can have excellent lab TBT and still feel sluggish to a real user clicking through it.
What field data is – and why it’s the verdict
Field data is what real visitors actually experienced, collected passively by Chrome and aggregated in CrUX over a rolling 28-day window. Google evaluates it at the 75th percentile, meaning at least 75% of real page views have to hit the good threshold for the page to pass. This is the data that feeds Google’s ranking signals, and it reflects the reality the lab can’t see: older phones, weaker networks, real interaction patterns, and the long tail of conditions your actual buyers browse under.
If you want the underlying benchmark those thresholds come from, what counts as a fast website for B2B SaaS lays it out. The short version: field data is the scoreboard. The lab score is practice.
Why a 100 can still be slow in the real world
When a perfect lab score and a failing field assessment sit side by side, the gap usually comes from one of these:
Real devices are slower than the simulation. Your buyers aren’t all on recent phones with strong signal. The 75th percentile deliberately captures the slower end of that range.
Real interactions surface INP problems TBT missed. The lab never clicks your nav, opens your menu, or submits your form. Real users do, and that’s where responsiveness failures appear.
Scripts that load conditionally. Chat widgets, A/B tools, and marketing tags that don’t fire in a clean lab run absolutely fire for real visitors – and compete for the main thread.
Geography and caching. A visitor far from your server or hitting an uncached page has a different experience than a single test from one location.
None of these show up in the number you were celebrating.
The flip side: when you have no field data
There’s a second trap, and new SaaS sites hit it constantly. If your site doesn’t yet have enough traffic, CrUX has nothing to report, and PageSpeed Insights will say there’s not enough field data – leaving you with only the lab score. That’s not a problem, as long as you read it correctly: the lab score is a reasonable proxy while you wait for real data, not a guarantee. The right move is to build in margin – aim well inside the thresholds rather than just clearing them – so that when real visitors do arrive, their slower-than-lab experience still lands in the green.
So what should you actually look at?
A simple hierarchy of trust:
Read the Core Web Vitals Assessment first – Passed or Failed, the field verdict. That’s your real grade.
Use the lab score as a tool, not a trophy – it’s how you diagnose and validate fixes, not how you measure success.
Check Search Console’s Core Web Vitals report – it groups field data by template, so you see which page type is failing, not just one URL.
Test on a real mid-range phone – the cheapest reality check there is.
Optimizing for a 100 in the lab while the field assessment reads “Failed” is polishing the practice score while losing the game. If your field data is failing and you’re not sure why, a website performance audit reads both sources and tells you the actual cause – and fixing it is exactly what speed optimization is for.
Frequently Asked Questions (FAQs)
It’s a good lab result – your page performed well in one simulated test. But it isn’t proof your site is fast for real users. Google ranks on field data (real visitor experience at the 75th percentile), which can fail even when the lab score is perfect.
Because the 0–100 score is lab data and the Core Web Vitals Assessment is field data. Real users on slower devices and networks, real interactions that surface INP problems, and conditionally loaded scripts all create a gap the lab test never sees.
Lab data is a single synthetic test from one simulated device – repeatable and good for diagnosis. Field data is what real visitors actually experienced over the last 28 days, aggregated by Google. Only field data determines your rankings.
No. Google ranks on the Core Web Vitals field assessment, not the 0–100 Lighthouse score. The lab score is a diagnostic tool that helps you improve the field metrics – it isn’t itself a ranking factor.
Your site doesn’t yet have enough traffic for Google to report real-user data in CrUX. Until it does, you only have the lab score. Treat it as a useful proxy, and build in performance margin so real visitors land comfortably inside the thresholds when they arrive.
Find out how fast your site really is
A lab score won’t tell you. Get a free 48-hour audit of your site – we’ll read your real field data, find where the gap is, and tell you exactly what to fix first.
