What a website audit should actually cover

An audit is worth what its first item is worth: a short, ranked list of defects that tells you what to fix before anything else.

Buy a website audit and you are often buying a mystery: nobody can tell you what will be looked at, the report arrives heavy as a brick, and somewhere inside it there is presumably a verdict. This vagueness is not tradition, it is a market failing, and you can refuse it. A proper audit is a structured look at a short list of questions, each of which a business owner would ask personally if the language were plain.

Here is what the structure covers. Do the pages load fast enough on a phone, where most of your visitors actually are. Does the site hold up on the screens your buyers use, not just the screen of whoever built it. Can engines find the pages and read them, so the site has a chance of appearing when someone searches for what you sell. Do the forms actually deliver, every enquiry landing somewhere a person reads it. And does the content still match what the business sells today, because sites drift out of date in ways their owners stop seeing.

Just as important is what an audit is not. It is not a magic ranking injection: nobody buys a position in results with a report. And it is not a sales document dressed as findings, where every defect happens to be something the auditor sells. When the findings map suspiciously neatly onto one vendor's service list, you have learned what the audit really was, and it was not an audit.

The failure versions cluster into three. The audit delivered as fifty screenshots from automated tools, with no ranking of what matters, so the reader cannot tell a fatal defect from a cosmetic one. The audit whose problems turn out to be the tool's upsell list rather than the site's. And the business that commissioned audits year after year and fixed nothing, then received the same findings twice, which is the moment an audit becomes an expensive way to learn nothing.

The practice is to interrogate the audit before you buy it. Ask what it covers, question by question, and walk away from anyone who cannot answer. Demand the ranked short list with evidence you can verify: screenshots of your own site failing at the moment it fails, timings you can reproduce, not generic scores detached from anything you can see. Then give one person the job of closing the items, because a list without an owner is a document, and documents decay.

We audit sites the way this essay describes, and we will happily tell a prospect their site does not need us, which is what a ranked list sometimes proves. Guidance, not guarantees: no audit can promise where a site will rank, because that lives with the engines and their changing machinery. What an audit can promise is an honest picture of what is broken, ordered by what it costs you, and a first item worth starting on Monday.