What a website audit actually measures — and what it cannot

Whether a title tag exists, whether it is unique, whether a canonical is set, whether robots.txt blocks the site, whether the page is served over HTTPS — these are facts. They are either true or false, and a tool reporting them is reporting, not estimating.

Load performance belongs here too, provided the numbers come from a real measurement source such as Google PageSpeed Insights rather than an invented score.

Whether your call to action is compelling. Whether your content is high quality. Whether your navigation is intuitive. Markup can hint at these — a page with no link whose wording invites action almost certainly has a conversion problem — but it cannot judge them.

A tool presenting a heuristic with the same confidence as a measurement is misleading you, even when its guess happens to be right. That is why every estimate in our reports carries a label saying so.

Any promise about rankings. Nobody outside a search company can guarantee position, and the ones who could do not sell audit tools.

Any promise about appearing inside ChatGPT or Google AI Overviews. What you can genuinely check is readiness: whether your content is structured, crawlable and explicit enough for a machine to quote accurately. That is worth doing. It is not the same as a guarantee, and conflating the two is a sales tactic.

If you cannot tell which findings are measurements and which are guesses, you cannot prioritise. You end up giving equal weight to a confirmed missing meta description and a speculative opinion about your colour scheme.

A score is only useful if you can see the reasoning behind every point it deducted. Ours is built that way: a category score is exactly 100 minus the sum of the issues listed, and each issue states what it cost.

That also means the number is reproducible. Run the same audit twice on an unchanged site and you get the same score, which is what makes it worth tracking after you make a change.

A single number is a poor description of a website and a good way to answer one specific question: did the thing I just changed help?

Used that way it is genuinely valuable. Audit, fix the top two issues, audit again, and the movement tells you whether your effort landed. Used as a vanity metric to be maximised, it encourages chasing trivial deductions while a slow homepage goes unfixed.

This is why ranking issues by their weighted effect on the total matters more than the total itself. The list tells you what to do; the number only tells you whether it worked.

Where does your performance data come from? If the answer is not a named source such as Google PageSpeed Insights or real field data, treat the numbers as decorative.

Which of these findings are measured and which are inferred? A tool that cannot answer this has not thought about it, and you will not be able to tell either.

Why did I lose those points specifically? If the deductions do not add up to the score, the score is not really derived from the findings, and neither is worth much.