The Site Audit Strategy Library
Three strategies for turning a crawl into scheduled work: triage the findings by severity, write the report so it gets approved, and decide which pages should stop existing. Each one includes a workflow and a copy-paste OpenSEO MCP prompt.
How do you run a site audit that ends in fixes?
Sort the findings by severity, decide which are worth someone’s sprint, and write the case so the work gets authorised. Deleting pages is a separate decision that needs its own evidence.
The technical SEO audit checklist that ends in fixes
One crawl returned 1,180 findings and 35 that mattered. Sort by severity, group by cause, and read the fix that ships with every issue.
02Write an audit report the client will actually act on
A six-section structure that ties every finding to a page, a cost, and a business number, plus what to leave out.
03Index bloat: when the fix is deleting pages
Five million pages came out of one site and it recovered. On a small site the same instinct usually wastes a weekend. How to tell which you have.
Why most audits stop one step short
A crawler is good at finding problems and has no opinion about which of them matter to your business. So the output is a list, the list is long, and the expensive work of deciding what to do with it gets deferred until nobody remembers why the crawl was run. Audits do not fail because the findings are wrong. They fail because a correct description of 1,180 problems gives the person paying for it no way to decide anything.
Two habits close most of that gap. Group findings by their underlying cause rather than by URL, because a templated site turns one mistake into hundreds of rows. And carry the remediation with the finding, so the person reading the report is not sent off to search for what a canonical conflict is before they can act.
What the OpenSEO site audit checks
The crawler is robots.txt-aware and stays on the same origin. It checks 27 issue types across three severities: four critical (a blocked crawler, a 5xx error, a broken internal link, a missing title), 14 warnings covering duplicate titles and descriptions, duplicate content, missing or multiple H1s, redirect chains and loops, canonical conflicts, thin content, missing image alt text, orphan pages and dead ends, and nine informational checks for length, heading order, response time, and intentional noindex or canonical signals. Every issue carries a how_to_fix written for that issue type.
You can run all of it from the site audit page, or through the OpenSEO MCP, which lets a compatible AI assistant start the crawl, poll it, read the issues with their fixes, and cross-check individual URLs against Google Search Console in one conversation. The seo-audit agent skill packages the same steps as a reusable command.
Free accounts can crawl 50 pages per audit, which covers most brochure sites. Larger crawls run to 10,000 pages on the $10/month plan. Lighthouse is optional and samples up to 10 representative pages rather than every URL.
Where the audit needs data the crawl does not have
Two of the most valuable audit outcomes are decisions rather than defects, and no crawler will surface either as a row.
- Which broken pages anyone was ever going to visit. Severity is a property of the issue; value is a property of the page. Cross the critical list against your Search Console queries and pages and the fixes that matter separate from the ones that are merely correct.
- Whether a wall of near-identical URLs is in the index at all. A crawler reports 800 healthy pages; URL inspection tells you Google never fetched 795 of them.
- Whether a competitor’s advantage is technical or structural. Before rebuilding a template, check the domains that actually hold your results, because half of them are frequently directories you were never going to outrank.
Site audit FAQ
What is a technical SEO audit?
A check of whether search engines can reach, render, and understand your pages. It covers crawl access, status codes, canonical and indexability signals, titles and headings, internal linking, duplicate content, and response time. It comes before content and link work, because a content problem on a page Google cannot fetch is not the problem you have.
What should a technical SEO audit checklist produce?
A work order rather than a count. Group findings by issue type instead of by URL, keep only the issues that stop a page being reached or understood in front of the reader, and attach the specific remediation to each one. A report of 1,180 findings across 318 pages usually describes about a dozen underlying causes.
Why do SEO audits produce so many issues?
Because most checks run per page and most sites are templated, so one template fault multiplies by every page using it. Two hundred and seventy-nine pages missing a meta description is one template change, not 279 tasks. Grouping by cause is what turns the number back into work.
What is index bloat, and how do I know if I have it?
More URLs eligible for indexing than the site has distinct things to say: pagination, filter parameters, tag archives, and per-item permalinks. A crawler cannot tell you whether you have it, because every bloated page returns 200 and passes its own checks. Inspect a sample of the suspect URLs in Search Console instead. If they come back unknown to Google or canonicalised away, there is nothing to delete.
Why did my SEO crawler get blocked?
A bot-protection layer refused it, usually with a 403, a 429 rate limit, or a managed challenge, and usually from the CDN edge rather than your server. Googlebot is normally exempt because vendors verify it by reverse DNS; third-party crawlers are not. Until access is fixed, every other number in the audit covers only the pages that were served.
Is there a free SEO audit tool?
Partly. Google Search Console reports coverage and indexing for your own verified property at no cost, and it is more reliable than any third-party estimate for anything Google-specific. A crawler adds the on-page and internal-link picture Search Console does not give you. OpenSEO is open source and free to start, with 50-page crawls on the free plan; the paid plan is $10/month and raises the limit to 10,000 pages per crawl.
Run a site audit with your own agent
Each strategy ends with a copy-paste MCP prompt. OpenSEO is open source, free to start, and does not require a credit card.