A Lighthouse SEO score of 100 is one of the most misleading numbers in web development. It is real, it is reproducible, and it says almost nothing about whether a page will rank.
That is not because Lighthouse is bad. It is because the SEO category is deliberately narrow. It checks a small set of machine-detectable signals and scores them. Everything that requires judgement, such as content quality, search intent, and links, is outside its scope.
Understanding what the score includes is the fastest way to stop over-optimising it.
What Lighthouse actually is
Lighthouse is an open-source auditing tool maintained by Google. It loads a page in a controlled Chrome instance, collects signals, and produces a report across four categories: performance, accessibility, best practices, and SEO. Each category has a score from 0 to 100, calculated from weighted audits.
It is a lab tool. It runs on one machine, at one moment, with one simulated network and device profile. That makes it excellent for catching regressions and comparing builds. It also makes it a poor proxy for what real users experience.
Run the checks yourself
Paste your URL and get the audit, not a checklist.
Launchkit checks metadata, canonicals, structured data, and Lighthouse in one pass.
Which audits feed the SEO score
The SEO category is smaller than most people assume. It checks whether a search engine can read and understand the page, not whether the page deserves to rank. The audits typically include:
- A valid
robots.txtthat does not block the page. - A
<title>element with meaningful text. - A meta description with meaningful text.
- A viewport meta tag that allows zooming.
- A valid
hreflangset, when present. - A valid
canonicallink, when present. - Crawlable links with descriptive anchor text.
- A valid
langattribute on the document. - A successful HTTP status code.
- Indexable content that is not blocked by
noindex.
Notice what is absent: internal link architecture, duplicate content across pages, content depth, structured data quality, and anything about the query the page is targeting. Lighthouse evaluates one URL in isolation. It cannot know that the same title appears on forty other pages.
Why a 100 is not a ranking guarantee
Ranking is a relative judgement. Search engines compare pages against each other for a query. A page can pass every automated SEO check and still lose to a competitor with clearer intent match, better internal links, and more authority.
A 100 means the page has no detected technical blockers in that category. It does not mean the page is useful, unique, or the best answer. It is a hygiene score, not a quality score.
The inverse is also true. A page can score 92 because of a missing meta description and still rank well, because the missing description is a snippet problem, not a ranking problem. Fix it, but do not treat the two-point drop as an emergency.
Lab data versus field data
Lighthouse produces lab data: a synthetic test under fixed conditions. Field data comes from real users, collected by the Chrome User Experience Report or your own analytics. They answer different questions.
Lab data is good for debugging a specific cause. If the largest contentful paint is slow in the lab, you can inspect the waterfall and find the image that is blocking it. Field data is good for knowing whether users are actually affected, and it is the data used for Core Web Vitals assessment.
The two often disagree. A page can be fast in the lab and slow in the field because of third-party scripts, slow APIs, or devices the lab profile does not represent. When they disagree, trust the field data for prioritisation and use the lab to diagnose.
Reading the other categories
The SEO score is only one quarter of the report. The other categories matter, sometimes more.
Performance measures loading and responsiveness: largest contentful paint, total blocking time, cumulative layout shift, and speed index. It is the category most likely to correlate with real user frustration. A slow page loses conversions before it loses rankings.
Accessibility checks contrast, labels, heading order, and ARIA usage. Some of it overlaps with SEO, especially semantic structure, but it is primarily about people. Treat it as a product requirement, not an SEO task.
Best practices covers security, console errors, deprecated APIs, and image aspect ratios. It is a grab bag. Some audits are important, such as HTTPS, and some are minor.
A page with 100 SEO and 40 performance is not in good shape. Read the whole report, not the number you were asked about.
What Lighthouse cannot see
These are the things worth checking manually, because no single-page lab tool can:
- Duplicate titles and descriptions across the site. Crawl the site and group by title.
- Canonical consistency. Compare the canonical, internal links, and sitemap for the same URL.
- Redirect chains and loops. Lighthouse follows redirects; it does not flag the cost of a long chain.
- Internal linking. Whether important pages are reachable and linked with useful anchors.
- Content quality and intent match. Whether the page answers the query better than the current results.
- Structured data accuracy. Whether the markup matches visible content.
- Indexation state. Whether the page is actually indexed, which requires Search Console, not Lighthouse.
- Crawl budget waste. Whether bots spend time on parameterised or duplicate URLs.
- Mobile parity. Whether the mobile DOM contains the same content and links.
- Rendering. Whether the primary content exists without JavaScript.
Run a crawl and check these directly. A site-wide crawl answers questions a single-page audit cannot.
A sensible workflow
Use Lighthouse as a regression check, not a target.
- Pick the templates that matter: homepage, a product page, an article, and a conversion page.
- Run Lighthouse on each, on a production build, with a clean browser profile.
- Fix the SEO audits that indicate a real blocker, such as
noindexor a blocked page. - Fix performance regressions that affect the key templates, measured against your own baseline.
- Run a site-wide crawl for the issues Lighthouse cannot see.
- Re-check field data after a week or two, not immediately.
| Question | Tool |
|---|---|
| Is this page blocked or noindexed? | Lighthouse, Search Console |
| Are titles duplicated across the site? | Site crawl |
| Is the page indexed? | Search Console |
| Are real users experiencing layout shift? | Field data |
| Is the canonical consistent? | Site crawl |
| Is the content the best answer? | Human judgement |
The last row is the one that decides rankings, and it is the one no score can measure. Lighthouse tells you the page is technically legible. It cannot tell you the page is worth reading.
If you want the crawl and the audit in one pass, run a free audit on a site you own. It combines a site crawl with a real Lighthouse run, so you see both the automated score and the issues the score misses. For the broader checklist, read the technical SEO audit guide, and see the privacy policy for how audit data is handled.