You run a Lighthouse audit, get a number, and… now what? A 74 — is that good? What's dragging it down? And is chasing a perfect 100 worth it? Lighthouse is one of the most useful tools in web performance, but only if you know how to read it. Let's decode the score.
What Lighthouse is
Lighthouse is an open-source auditing tool (built into Chrome DevTools and many services) that loads your page in a controlled environment and grades it across several categories, each on a 0–100 scale:
| Category | Measures |
|---|---|
| Performance | How fast the page loads and responds |
| Accessibility | How usable it is for people with disabilities |
| Best Practices | Security and modern web standards |
| SEO | Whether search engines can crawl and understand it |
The Performance score is the one people obsess over, but the others are quietly valuable too.
How the Performance score works
The Performance number isn't one measurement — it's a weighted blend of several metrics, including lab versions of the Core Web Vitals:
- LCP — when the main content appears.
- CLS — how much the layout shifts.
- Total Blocking Time — a lab proxy for responsiveness (INP).
- First Contentful Paint and Speed Index — early rendering milestones.
So a low score points you at which of these to fix — the report breaks it down for you.
Lab data — its strength and its catch
Here's the crucial thing to understand: Lighthouse is a lab test. It runs once, on one simulated device and network. That makes it:
- ✅ Great for debugging — consistent, repeatable, with specific recommendations.
- ✅ Great for catching regressions before release.
- ⚠️ Not the same as real users. Your actual visitors are on many devices and networks — and Google ranks on field data, not your lab score.
Don't confuse your Lighthouse score with what users experience. Lab is for finding and fixing problems; field data (real visits) is the truth for ranking. Use Lighthouse to improve, then confirm with real-user data.
The trap of chasing 100
A perfect 100 is tempting, but it has sharply diminishing returns:
- The jump from 50 to 90 usually comes from a few high-impact fixes (images, caching, JS) and is hugely worth it.
- The grind from 95 to 100 can eat days for gains nobody perceives.
Aim for a solid 90+ and a good real-world experience — not a vanity 100. Past a point, your time is better spent elsewhere.
How to use it well
- Run it on your key pages (not just the homepage).
- Read the "Opportunities" and "Diagnostics" — they're a prioritised to-do list.
- Fix the biggest items first — usually images, render-blocking resources, and TTFB.
- Re-run to confirm each change helped.
- Don't stop at the lab — verify with field data, and monitor performance over time so a regression doesn't sneak back.
The bottom line
| In one line | |
|---|---|
| What | A 0–100 audit across performance, a11y, best practices, SEO. |
| Performance score | A weighted blend of lab Core Web Vitals + render metrics. |
| Strength | Great for debugging and catching regressions. |
| Catch | It's lab, not field — aim for 90+, not a vanity 100. |
Lighthouse is a fantastic flashlight for finding what to fix — just remember it's a lab, not your users. Use it to climb to a strong 90+, then trust real-world data for the final word.
Test any page (with a Lighthouse-style audit) using the free Core Web Vitals checker, and track scores over time with performance monitoring.