Audit · 6 min read · 16 May 2026
AI website audit: what it actually checks
An AI website audit checks the things a human reviewer cannot read at scale: structured data, machine-readable intent, AEO signals, and where revenue may be leaking. Here is what it covers and what it does not.
An AI website audit is not a redesign brief. It is a structured read of the website from the angle that humans cannot scale: how cleanly the site communicates intent to search engines, AI answer engines, and ad platforms, and where the underlying systems are quietly losing trust, traffic or revenue.
The point of running one before any rebuild work is simple. Most stores do not have a design problem. They have a clarity, structure and tracking problem. An audit decides whether a rebuild is even justified.
What an AI website audit checks
A serious audit covers four layers.
The first is clarity and intent. The audit reads the page the way a language model reads it: what is the page about, who is it for, what action does it expect. If the answer is not obvious in the first 200 words, neither AI assistants nor first-time buyers will understand the offer.
The second is machine readability. This is where most websites lose. The audit checks semantic HTML, heading hierarchy, schema markup (Organization, Service, Product, FAQPage, BreadcrumbList, Article), Open Graph metadata, canonical URLs, robots and sitemap discipline, and the presence and quality of an llms.txt summary for AI assistants.
The third is trust and conversion signals. Reviews, named humans behind the business, real engagement scope, legal pages, contact paths, and how confidently the offer is described. Premium buyers and AI summarisers both penalise vague pages.
The fourth is operational health. Tracking accuracy, recovery flows, app dependency, and where the data actually lives. A polished front-end on a broken backend leaks revenue silently.
What it does not check
An AI website audit does not guarantee rankings, AI recommendations, or revenue. It identifies the issues that may be affecting clarity, trust, machine readability, conversion flow and tracking, and prioritises them by impact and effort. The fix is still a human decision.
It also does not replace a security or accessibility audit. Those are separate disciplines with their own checklists.
Why this matters now
Search has split. Some buyers still type into Google. A growing share now ask an AI assistant. Both routes depend on the same foundation: a site that can be read cleanly and understood quickly. A website that converts well on Google but is invisible to AI answer engines is missing half of modern intent.
The cheapest improvement most websites can make this year is not a redesign. It is becoming legible to the systems already trying to read them.
When to commission one
Three triggers usually justify an audit:
- The site is converting below where the brand, traffic and offer suggest it should.
- The team is considering a redesign and wants to know whether the design is actually the problem.
- The site needs to start showing up in AI assistants and answer engines and currently does not.
In each case, audit first. The audit defines the work. The rebuild is downstream.