Search for a way to check whether your website is ready for AI search and you will find a growing list of audit tools, many of them free. They all produce a score and a list of recommendations. They do not all measure the same things, and some measure very little. If you are going to act on a report, it helps to know what a good one looks like.
By the iSuggest.ai Team ยท Updated for 2026
This post is a buyer's checklist. It explains what a GEO audit can honestly tell you, which checks matter most, how to read a score without being misled, and which red flags suggest a tool is selling comfort rather than information. We run a free audit ourselves, so we have an interest here, and we apply the same checklist to it. We make no comparisons with other products. If you need background first, start with what GEO is.
What an audit can and cannot tell you
An audit is a set of tests run against your pages. It can tell you whether a crawler is able to fetch a page, whether the text is present in the HTML, whether structured data is valid, whether headings make sense, and whether identity signals such as an organisation name and author are clear. These are properties of your site, and they can be measured the same way each time.
What an audit cannot tell you is whether ChatGPT, Claude, Gemini or Perplexity will mention you tomorrow. Those systems are complex, they change, and their answers vary between runs. A tool that claims to predict or guarantee citations is overreaching. The honest framing is that an audit finds the reasons you might be skipped and helps you remove them. It improves your chances; it does not decide the outcome.
Keep that framing in mind throughout. A good audit is a diagnostic, not a promise.
Why the choice of audit matters now
The channel is growing fast enough to justify the effort. Industry analysis reports that traffic referred by AI search grew about 16x between 2024 and 2026 (SEO Sherpa), and a compilation reports roughly 68% of US Google queries end without a click (Digital Applied). Those are industry figures with their own methods, and growth from a small base is still small, but the direction is clear. Meanwhile, a study reported by LLM Pulse found that over 35% of the top 1,000 sites block at least one major AI crawler, often without realising. An audit that catches problems like that is genuinely useful. One that misses them is worse than nothing, because it gives false reassurance.
Checklist item 1: It tests access first
The first question any audit should answer is whether AI crawlers can reach your pages at all. If they cannot, nothing else matters. A good tool checks several layers.
- robots.txt rules for the named AI crawlers, not only for a generic wildcard.
- Meta and header directives such as noindex or nosnippet that restrict use of a page.
- HTTP status and redirects, so you know a crawler gets the page and not an error or a loop.
- Canonical tags that point to the version you want read.
Be wary of a tool that never mentions crawler access. Our guides on checking whether GPTBot can crawl your site and robots.txt syntax mistakes show how much can go wrong here, and the rise of AI crawlers explains who the crawlers are.
Checklist item 2: It looks at the HTML a crawler receives
Many AI crawlers do not run JavaScript. If your page builds its content in the browser, a crawler may receive a nearly empty shell. A good audit fetches the raw HTML, the way a simple crawler would, and checks whether the main text, headings and links are in it. A tool that only inspects a fully rendered page in a browser can miss the problem entirely, because everything looks fine to a person.
Ask of any tool: does it say which content is present before scripts run? Our post on what client-side rendering hides from AI crawlers explains why this single check explains so many mysteries.
Checklist item 3: It validates structured data, not just detects it
Structured data gives machines a labelled version of your facts. A shallow audit says "structured data found". A better one tells you which types are present, whether required fields are filled, whether it parses without errors, and whether it matches what is visible on the page. Markup that describes things not on the page is a risk, and a good tool flags mismatches.
Look for specific, actionable findings: "Organization markup is missing a name" is useful; "improve your schema" is not. For the reasoning behind this, see structured data for AI search engines and FAQ schema.
Checklist item 4: It examines content structure
Models and retrieval systems extract passages. Pages that are easy to extract from tend to share traits: one clear H1, descriptive headings, short paragraphs that lead with the answer, lists and tables where the content is list-like, and a visible date. A good audit looks at these. It should not try to grade writing quality with a single magic number, but it can reliably flag structural problems such as skipped heading levels, empty headings or walls of text.
Our guide to writing content AI models want to quote describes the target shape.
Checklist item 5: It checks identity and trust signals
For a system to recommend a business, it must be confident about who the business is. Useful checks include a clear organisation name, consistent contact details, an About page, named authors with real credentials where relevant, and links to official profiles. A tool that ignores these is missing the part that explains why two otherwise similar sites get treated differently. See entity clarity for why this matters.
Checklist item 6: It is transparent about what it measures
You should be able to find out what a score is made of. Look for a plain list of the checks, what each one tests, and how it contributes. If a tool hands you a single number with no breakdown, you cannot tell what to fix or whether the number is meaningful. Transparency also lets you judge whether the tool's priorities match your situation. Ours is described in inside an iSuggest.ai audit.
Checklist item 7: It is repeatable
Run the same audit twice on the same unchanged page. You should get the same result. If the score wobbles for no reason, you cannot use it to track improvement, because you will not know whether a change in the number reflects a change in the site. Repeatability is what turns an audit into a tool for tracking progress. A page's score should move when you fix something and stay put when you do not.
This is also why audits sit well alongside manual checks. Asking an assistant real questions is valuable, but its answers vary run to run, so it cannot serve as a stable measurement. A deterministic audit measures your site; manual prompting samples the assistants. You want both, and you should not confuse them.
Checklist item 8: It gives prioritised, specific fixes
A long list of 60 equal-looking recommendations is exhausting and rarely acted on. A good report ranks issues by impact and effort, says exactly what to change and where, and separates blockers from polish. The ideal first screen answers three questions: what is stopping me, what is the quickest win, and what can wait.
Vague advice is a warning sign. "Add more content" and "improve authority" are not fixes. "Your robots.txt blocks ClaudeBot on line 12" is.
Checklist item 9: It is honest about cost
"Free" can mean several things. It might mean free forever, free for one page, free with an email wall, or free for a trial. Check what you actually get: how many pages, whether you can re-run it, whether results are saved, and what is held back for paid tiers. There is nothing wrong with paid features, but you should know where the line is before you invest time.
For transparency on our side: audits on iSuggest.ai are free and not metered, so you can run them as often as you like. Publishing a snapshot of your business to the AI directory is the part that spends credits, which are non-refundable; our post on how credits work explains the model.
Checklist item 10: It respects your data
An audit fetches your pages, so ask what it keeps. Does it store the content it reads? For how long? Is anything shared with third parties? Does it require an account to see results? A reputable tool states this plainly. If you cannot find a privacy statement, treat that as an answer.
Red flags
Guaranteed citations or rankings
No honest tool can guarantee that an AI system will cite you. Language like "get cited by ChatGPT in 30 days" is a promise nobody controls.
One impressive number with no detail
A big green "92" feels good and teaches you nothing. If you cannot see what is behind it, you cannot trust it.
Recommendations that are all "buy more of our product"
If every finding resolves into an upsell, the audit is a sales funnel. Good findings are things you can fix yourself.
Fake precision
Claims such as "your site is 73.4% likely to be cited" cannot be backed by anything a tool measures. Treat decimals on predictions as decoration.
No mention of AI crawlers
If an "AI audit" does not examine AI crawler access, it is an ordinary SEO audit with a new label.
Invented statistics
If a report quotes figures without a source, you cannot check them. Look for named sources and honest framing of what each number covers.
How to run a fair comparison on your own site
You do not need to take anyone's word for what a tool finds. A simple experiment works with any audit.
- Pick a page with a known problem. For example, a page you know blocks a crawler, or one that loads its text with scripts.
- Run the tool. Does it find the problem you planted? If not, it is missing something real.
- Fix the problem and re-run. Does the finding clear and the score move in the expected direction?
- Run it twice without changes. Is the result the same?
- Read the fixes. Could you act on each one today without asking anyone what it means?
A tool that passes these five steps is doing its job, whoever made it. One that fails should not be driving your priorities.
A worked example: reading one report
Imagine a small consultancy runs an audit on its homepage and gets a middling result with five findings. This is an illustration, not a case study. The findings are: a robots.txt rule that blocks one AI crawler by accident, key service text that only appears after a script runs, an Organization markup block missing its name, a page with two H1 headings, and no visible updated date.
A weak approach is to try to fix all five on the same afternoon without understanding them. A better approach is to order them. The robots.txt rule and the script-dependent text are blockers: if a crawler cannot fetch the page or see its text, nothing else counts. The missing organisation name, the double H1 and the date are quick polish. The consultancy fixes the blockers first, confirms with a re-run that those findings clear, then does the polish. The whole job takes an afternoon, and it can state exactly what changed.
The lesson is the order. A good audit makes that order obvious.
What to do with the report
An audit changes nothing by itself. Use it in a loop.
- Run it on the pages that matter most: homepage, main product or service pages, and your best content.
- Fix blockers first, then structure, then polish.
- Re-run to confirm each fix.
- Record the score and date so you can see progress.
- Repeat on a schedule. Our guide to how often to re-audit suggests a rhythm.
Then look at outcomes: referral traffic from assistants, and whether you appear when you ask the questions your customers ask. Our post on measuring AI search traffic in Google Analytics covers the first, and the AI SEO checklist gives a manual list to cross-check any report against.
Where our audit fits
The free iSuggest.ai audit is built around this checklist: access, raw HTML, structured data, content structure and identity signals, with a transparent list of what is measured and the same result each time on an unchanged page. It does not promise citations, and it does not need an account to be useful as a diagnostic. Judge it the way you would judge any other, using the five-step test above.
When your pages pass, you can publish a factual snapshot of your business to the iSuggest.ai AI directory, which has pages for ChatGPT, Claude and Gemini. It adds one more consistent, crawlable description of you and improves findability without guaranteeing a citation. See how it works, run an audit on your homepage, then sign in at your account page when you are ready.