Headline numbers
- 4,963 URLs crawled before the 5,000 cap, all of them HTTP 200, no fetch errors, no broken internal links
- 781,000 further URLs were discovered and not crawled; robots.txt kept the crawler out of 1,299 and 3,965 pointed at other gov.uk hosts
- 55 URLs redirect, 13 of them in chains of two or more hops, one of three
- 0 pages with a missing title; 19 indexable pages share a title with another; 252 indexable pages carry no canonical tag
- 411 pages have no meta description and 1,218 have one longer than 160 characters
- 1,282 titles run past 60 characters
- 1,066 guide pages carry two H1s, by design
- 3,356 images on the crawled pages, 21 of them without alt text, all the same licence logo
GOV.UK is the benchmark. It is the most audited public website in the country, built by a team that publishes its own standards, so we crawled it first to show what a clean site looks like inside Crawl Cove, and what is still left over when a site is that clean. Everything below is what the crawler saw on 29 October 2026 from a crawl capped at 5,000 URLs, which is a thin slice of a site that lists hundreds of thousands of pages. It is a reading of that slice, not a verdict on GOV.UK.
The crawl
4,963 URLs fetched before the cap; every one of them answered 200. No server errors, no client errors, no timeouts, no broken internal links. The crawler discovered another 781,000 URLs it did not fetch, which is the honest measure of how much of GOV.UK five thousand pages covers: the Overview's health score says so itself ("scores the 1% of discovered URLs this crawl reached"). robots.txt kept us out of 1,299 URLs, and 3,965 links pointed at other gov.uk hosts (assets.publishing.service.gov.uk alone accounted for 1,844), which a single-host crawl leaves alone.
What good looks like, in numbers: 0 missing titles across 4,902 HTML pages, 3,356 images with only 21 missing alt text, redirects on just 55 URLs out of 4,963, and a site that answered 4,963 requests at two a second without a single hiccup.
Five things the GOV.UK technical SEO audit noticed
1. Thirteen redirect chains, one of them three hops
What it is: 55 crawled URLs redirect. 42 are single hops; 13 take two or more, and one (a National Insurance statement page) takes three 301s to land on its current guidance URL. Three chains pass through an http:// URL on the way to https. Why it matters: each hop costs the visitor time and the crawler budget, and search engines follow only a few before they stop. On a site this size 13 chains is a rounding error, which is the point: they still exist. The free redirect checker runs the same check on a single URL.
2. Nineteen indexable pages sharing a title
What it is: 73 title strings are used by more than one crawled page, 490 pages in all, but the crawler raised only 19 pages as duplicates. The rest are search listing pages ("News and communications - GOV.UK" appears 78 times) that canonicalise to a single URL or carry noindex, so they are the same page to Google and the check leaves them alone. The 19 that remain are real pairs: a publication and the guidance page that replaced it, a consultation and its outcome. Why it matters: Google picks one page per title to rank and quietly suppresses the other. Fix: a distinct title for whichever of the pair is meant to rank.
3. Meta descriptions: 411 missing, 1,218 too long
What it is: of 4,902 HTML pages, 411 have no meta description, 1,218 have one over 160 characters and 666 have one under 70. Why it matters: Google writes its own snippet when the description is missing or cut, and GOV.UK's content is exactly the kind where the first sentence is a good snippet anyway, which is why this sits at medium, not high. The long ones are mostly publication summaries pasted whole into the tag.
4. 252 indexable pages with no canonical tag, 392 pointing elsewhere
What it is: 252 pages that can be indexed carry no rel=canonical at all, and 392 declare a canonical that is a different URL (the filtered search listings above, mostly, which is correct). Why it matters: a missing canonical is harmless until the same page is reachable at two URLs, and GOV.UK's redirect discipline means it rarely is. It is a belt-and-braces fix. The canonical issue page explains the check.
5. Two H1s on 1,066 guide pages, and 21 alt-less copies of one logo
What it is: GOV.UK's multi-part guides (Universal Credit, eVisas, HMRC sign-in) put the guide title and the part title both in an H1, so 1,066 pages report two. That is a design decision the crawler flags because most sites that have two H1s did not mean to; here it is deliberate and consistent, and we note it rather than score it. The 21 images without alt text are all one file: the Open Government Licence logo in publication footers. One template fix clears all 21. The image alt text checker runs the same check on one URL.
Also seen and left alone: 2,269 Article and 225 NewsArticle JSON-LD nodes without a headline property (Google lists headline as recommended, so this is a tidy-up), 1,282 titles over 60 characters (GOV.UK favours full descriptive titles and accepts the cut in the result), 66 noindex pages, and 651 near-duplicate pairs, almost all of them multi-part guides whose parts share the same frame.
What we'd fix first
- Point the internal links behind the 13 chained redirects at their final URL, starting with the three-hop National Insurance page.
- Give the 19 duplicate-title pairs distinct titles, choosing which of each pair should rank.
- Add alt text to the Open Government Licence logo in the publication template (clears all 21 image findings in one edit), then write meta descriptions for the 411 pages without one, starting with the most-visited.
That is the whole list for one of the best-run sites in the country, which is roughly what we expected. The crawl itself took 45 minutes at two requests a second.