If you have run a Google Lighthouse test on the same page twice in a row and got two different scores, you have not done anything wrong, and your site has not actually changed in the seconds between tests. This is genuinely normal behaviour, and understanding why it happens will save you a lot of confused re-testing.
This article picks up where our earlier piece on what Google Lighthouse is and how to use it left off, focusing on why scores fluctuate and on a related, often misunderstood point: why a good Google Lighthouse score isn’t automatically the same as good real-world performance in Google’s eyes.
Why the Same Page Scores Differently Each Time
- Google Lighthouse runs a single simulated page load, and network conditions, server response variability and even your own computer’s momentary workload can all shift the result slightly between runs
- Third-party scripts, adverts, tracking pixels, and embedded widgets often load unpredictably, and their exact timing can meaningfully affect the metrics Google Lighthouse records on any given run
- Testing conditions like throttling settings, the simulated device and network speed, are configurable, and a score run under one set of conditions is not directly comparable to a score run under another
- Server load at the exact moment of testing, particularly on shared or lower-tier hosting, can introduce genuine variability into response times
None of this means Google Lighthouse is unreliable. It means a single score is a snapshot, not a precise, fixed measurement, and it is worth treating it that way: run a test several times and look at the general range rather than fixating on one specific number.
Lab Data vs Field Data: The Distinction That Actually Matters
This is the part that trips up a lot of site owners, and it is worth understanding properly. Google Lighthouse produces what is known as lab data, a simulated test run under controlled, artificial conditions in one location, on one connection speed, at one moment in time.
Google’s actual ranking systems, however, lean heavily on field data, real measurements gathered from genuine visitors using your site in the real world, captured through the Chrome User Experience Report, often referred to as CrUX. This field data reflects a wide range of real devices, connection speeds, and locations, rather than one artificial test scenario.
It is entirely possible, and genuinely common, for a site to score well in a Google Lighthouse lab test while its real-world field data, what Google actually weighs for Core Web Vitals in search, tells a different, less flattering story, particularly if your real visitors are on slower connections or older devices than the ones Google Lighthouse simulates by default.
Where to Find Your Real-World Field Data
- Google Search Console’s Core Web Vitals report shows genuine field data gathered from your actual visitors, grouped by URL and issue type
- PageSpeed Insights displays both lab data from a live Google Lighthouse test and, where enough real-world traffic exists, field data drawn from CrUX, shown separately on the same report
- The CrUX dashboard and API offer a more detailed, historical view of field data for sites with sufficient traffic volume to generate a meaningful dataset
If your site does not receive enough traffic for CrUX to generate field data, Google generally relies more heavily on lab data as a reasonable proxy, which is one further reason smaller sites should not obsess unduly over chasing a perfect Google Lighthouse score in isolation.
How to Get a More Reliable Reading
☐ Run Google Lighthouse several times rather than relying on a single test, and look at the general range rather than one specific number
☐ Always check your Core Web Vitals field data in Search Console alongside any lab score, since this reflects what Google is actually measuring for ranking purposes
☐ Test with third-party scripts and adverts in their normal, live state, rather than in an artificially clean testing environment that does not reflect what real visitors experience
☐ Be cautious of any service or developer promising a fixed, guaranteed Google Lighthouse score, since some natural variability is simply how the tool works
☐ Treat a Google Lighthouse score as a diagnostic starting point for finding specific issues, not as the final, definitive measure of your site’s real-world performance
Frequently Asked Questions
Should I chase a perfect 100 Google Lighthouse score?
Not at the expense of everything else. A high score is a genuinely good sign, but real-world field data matters more for actual search performance. Chasing the last few points on a lab score, sometimes through fragile or overly aggressive optimisation, can occasionally introduce genuine usability trade-offs that are not worth the marginal score improvement.
Why does my site score well in Google Lighthouse but poorly in Search Console’s Core Web Vitals report?
This is precisely the lab versus field data gap. Google Lighthouse tests one simulated scenario, while Search Console reflects what your real, varied visitor base experiences, including different devices, connections, and locations. A gap between the two usually means real users, particularly those on slower connections or older devices, are having a noticeably worse experience than the lab test suggests.
Does mobile or desktop testing give a more accurate score?
Mobile testing is generally the more meaningful benchmark for most sites, since Google predominantly uses mobile-first indexing and most web traffic today is mobile. It is still worth checking both, since a site can occasionally have distinct issues on one that are not present on the other.
Can third-party scripts like Google Analytics or ad networks lower my score unfairly?
They can genuinely affect your score, but it wouldn’t be quite right to call it unfair, since those scripts add real load time that your actual visitors experience too. It is worth understanding which specific third-party scripts have the biggest impact, so you can decide what’s genuinely necessary versus what you can remove, delay, or replace with a lighter alternative.
How often should I check my Google Lighthouse score?
After any significant change to your site, a new plugin, a theme update, a design change, is a sensible trigger point. Beyond that, checking your Core Web Vitals field data in Search Console on a regular basis, monthly, is reasonable for most sites and gives a more meaningful, ongoing picture than repeatedly re-running individual Google Lighthouse tests.
Want Your Site’s Real-World Performance Checked?
If you would like a proper look at both your lab scores and your actual field data, and what specifically is holding your real-world performance back, get in touch for a straightforward conversation with no obligation and no sales pitch.