An SEO audit that ends in decisions, not a 200-item list
You want to audit a site without producing a document that nobody acts on.
Short answer
Audit in order of blast radius: can pages be crawled, can they be indexed, do they answer a real query, and do they convert. Most published checklists invert this, opening with dozens of cosmetic checks. A useful audit ends with a short ranked list of pages to fix, not a catalogue of findings.
Level one: can it be reached
robots.txt, the status codes on your important pages, redirect chains, and whether the content is in the HTML or assembled in the browser. Anything failing here makes everything below it irrelevant, which is why it goes first and takes twenty minutes.
Level two: can it be indexed
noindex tags, canonicals pointing somewhere unexpected, duplicate hosts serving the same content, and pages missing from the sitemap or missing from your internal links. This is where most silent losses live, because every symptom looks like "we just are not ranking".
Level three: does anyone want it
Impressions by page. A page with none has no demand or has never been found, and no on-page work fixes either. This one check reclassifies a large share of any audit’s findings as not worth doing, which is the most valuable thing an audit can do.
Level four: is the result competitive
For pages with impressions and few clicks, read your own snippet against the results page. Title truncation, a description that does not match the query, a competitor with a richer result. This is the cheapest traffic available on any site and it is almost never in an audit template.
Level five: does it work when they arrive
Speed on real devices, interface failures, dead clicks, and whether the page delivers what the snippet promised. Traffic already acquired and lost is more expensive than traffic never acquired.
What to skip
Keyword density. Meta keywords. Word-count targets. Heading counts. Any score out of a hundred. These generate findings without generating decisions, and their main function is to make an audit look thorough.
The output
A ranked list of pages, each with the specific change and the reason. If a finding does not name a page and an action, it does not belong in the document. An audit that lists two hundred issues and no priorities has moved the work of deciding onto the reader.
Questions
- How long should an audit take?
- A first pass on a small site is a day, and most of the value is in the first two hours. Multi-week audits usually reflect the size of the template rather than the size of the site.
- Should I use an automated audit tool?
- For crawling and error detection, yes, they are much faster than doing it by hand. For prioritisation, no: they flag everything equally because they cannot tell which pages matter to you.
- How often should I audit?
- After a migration, after a traffic change you cannot explain, and otherwise once or twice a year. Continuous auditing produces a backlog of low-value findings that competes with actually building pages.
Measured, not asserted
Level three is where most audits should stop for most pages. On ihatepdf.cv, 93 of 160 pages earned fewer than ten clicks across 224 days, and 22 earned none: auditing the on-page details of those pages would have produced findings for pages that had no demand behind them in the first place.
Free tool for this: SERP Audit. No account, nothing uploaded.
Where this goes deeper
Every number on this page comes from one complete dataset: one product taken from zero to 100K+ users on search alone, with nothing spent on advertising. The full argument is Chapter 30 of the book. Five chapters are free to read.