Free SEO audit tool

Everything here is measured from outside your site, which is the reason it works on any domain and the reason it has blind spots worth naming. It does not know your impressions, your average position, or which queries you already win. Search Console owns that data and nothing can infer it from the HTML.

It has no keyword data either. A page can pass every check on this list and still target a phrase nobody searches, or compete for one where the first page is owned by sites with twenty years of references. Passing means the page is not being held back by anything mechanical, not that the topic was worth writing about.

Off-site signals are absent entirely: who links to you, who mentions you, what your brand looks like to a search engine that has seen the rest of the web. That gap is the honest reason two sites with identical audit scores rank differently. What this audit does own is the half you control, and on most sites the mechanical half is where the recoverable losses are: pages that cannot be crawled or indexed and pages too thin to deserve the traffic.

Take crawl access first. If robots.txt or the sitemap is wrong, every other number in the report was measured over the wrong set of pages, and fixing anything else is guesswork. This is the layer where one line of text can hide an entire site.

Indexing signals come second, because they decide which of your URLs a search engine treats as canonical. Self-referential canonicals, honest titles and descriptions and clean redirects usually settle most of what a duplicate report is complaining about, without a word of content being rewritten.

Then work templates before pages. A missing description on 200 URLs is one template bug, not 200 tasks, and the same is true of duplicate titles, missing image dimensions and unsized hero elements. Only after that does per-page work pay: merging the thin pages, deepening the ones worth deepening, and leaving alone the pages that were always meant to be short.

Re-run the audit between layers rather than at the end. One template change routinely clears dozens of individual warnings, and the second run tells you which of the remaining items were real rather than downstream of something you have already fixed.

The report header on a typical small-business site reads: 34 pages crawled, 27 listed in the sitemap. That mismatch is the first finding, before any check has been read. Seven pages are reachable by links but unlisted, usually fine, or listed but orphaned, worth knowing. Neither number is the site's real size, and every ratio below is computed over the 34 actually fetched, which is why the header is worth reading before the scores.

The technical column then shows canonical coverage at 62%, against the 80% pass line. Clicking through shows the pattern in one glance: every page missing a canonical is an article, every page carrying one is not, so the article template is the single bug behind thirteen warnings. The same column shows three redirect chains from an old http-to-www migration and two templates past the load ceiling, both of them page types rather than pages.

The content column reports nine pages under the 300-word floor and two duplicate pairs at 0.61 and 0.58. The pairs turn out to be print-friendly variants the theme generates, a canonical fix, not a writing task, and six of the nine short pages are event announcements from past years that should be archived rather than expanded. What looked like eleven content problems is one template setting and one editorial decision.

That compression is the pattern to expect: 30-odd findings usually reduce to four or five underlying causes, and the report is designed to be read that way, grouped, not item by item. Fix the causes, re-run, and treat whatever survives the second crawl as the real per-page work. When the survivors are mostly content items, the content audit reads those pages more deeply than this preset does; when they are crawl-and-indexing items, the technical audit is the faster loop to iterate with.