I audited my own website six ways. Here's what it found.

Why run six tools instead of one?

Because every audit tool has opinions. One will give a page 55 and another 78 for the same thing. Running two per area and averaging them smooths out the quirks, and where both agree on a fault, it's almost certainly real. All six are free and run locally with no sign-up.

What were the scores?

Averaged across both tools in each column, noodell.com scored 95 for SEO, 66 for GEO and 66 for AEO, out of 100. The home page scored well on everything. The inner pages dragged every column down, most of all on the two strictest tools.

Why did a 95 for SEO mean so little?

The SEO tools only check the hygiene of pages that exist: tags, speed, links. They can't see that the site ranked in the top eight on Google UK for none of the nine searches I actually care about. Seven tidy pages can't rank for questions they never answer.

What did the audits find?

Ten faults worth fixing. The worst was on the free tool page itself: its page header closed on the first line, so its title and description sat where Google doesn't reliably read them. Five of the six tools flagged it. My best page for "free process cost calculator" wasn't saying what it was.

• Only the home page had structured data, link-preview tags or a canonical address. Every inner page had none.
• No question-led content: one heading in twelve was a question, and there were no FAQs or short direct answers for AI tools to quote.
• Web fonts blocked rendering for 2.3 seconds, pushing first paint on mobile to 3.5 seconds.
• An internal note left in one page's source was visible to every AI crawler, and one tool flagged it as a prompt-injection pattern.
• No dates, authors or RSS feed anywhere but the home page, so nothing signalled freshness or who wrote it.

What's changing?

Every existing page is being brought up to the home page's standard in one pass, with no new copy needed. Then the site grows from seven pages to about twenty, each aimed at a real search: a Glasgow page, a guide to costing a process, a page for each service, and these field notes.

The first of those is live: how to work out what a manual process costs your business. The scores get re-run after every change, and I'll post the before and after here.

Re-run scores after the fixes go live: to follow.

What should you take from it?

A high score from one tool proves very little. Run two per area, believe the faults they agree on, and check whether you actually appear for the searches your customers type. If you don't, the answer is usually pages that answer real questions, not more tags on the pages you have.

It's the same lesson as process mapping: you can't fix what you haven't measured. Measure one process in ten minutes, free.