Free AEO Checker: What an Honest Answer Engine Check Shows

An answer engine is any AI product that replies with a written answer instead of a list of links: ChatGPT, Perplexity, Claude, Gemini, and Google's AI Overviews and AI Mode. Answer engine optimization is the work of being named, and named correctly, in those answers. An honest check looks at four things.

• Mention. For a question a buyer would actually ask, does the engine name you at all?
• Accuracy. When it names you, are the facts right? Prices, product names, location and what you sell are the usual failures.
• Citations. Which URLs does it cite? If it cites review sites, directories or competitors, those pages are shaping the answer.
• Access. Can each engine's crawler fetch your pages, and is your key content in the HTML it receives?

The first three describe the answer. The fourth is the part you control most directly. For a fuller breakdown of each signal, see what an AI brand checker actually checks.

Many AEO tools grade you on FAQ schema, answer-box formatting, heading structure or "AI-friendly" writing. The evidence doesn't support any of those as levers for AI answers.

• Google's AI optimization guide says you don't need new markup, machine-readable files or a special writing style to appear in its generative AI features.
• Ahrefs tracked 1,885 pages that added schema against matched controls and found no major uplift on any platform, with AI Overview citations falling 4.6%.
• Google deprecated FAQ rich results in May 2026.
• Google rejects chunking content for LLMs and says there is no ideal page length.

Don't overcorrect. Google says existing markup does no harm, so there's no reason to rip it out. It just isn't a lever for AI answers. A checker that hands out points for markup is measuring what it can count, not what the engines are shown to reward.

1. Write three buyer questions. Use questions a buyer asks before they know your name, such as "best payroll software for a 20-person agency", not your brand name.
2. Ask each engine logged out, or with memory off. Use ChatGPT with search, Perplexity, and a Google search that shows an AI Overview. OpenAI says ChatGPT may use saved memories when rewriting a search query, so a logged-in test reflects you, not a buyer.
3. Record each answer. One spreadsheet row per question per engine: named or not, the facts stated, and every cited URL.
4. Read your robots.txt at yoursite.com/robots.txt. Check the rules for the retrieval bots: OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot. GPTBot governs training only, and OpenAI says each setting is independent. This is valid:

User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

5. Test the firewall. OpenAI says eligibility means allowing OAI-SearchBot and confirming your host or CDN allows traffic from its published IP addresses. A rough first test:

curl -I -A "OAI-SearchBot" https://yoursite.com/

A 403 or a challenge page is a red flag. A firewall that checks IP addresses may treat your request differently from the real crawler, so your CDN's bot logs are the ground truth.

6. Check the raw HTML. Run curl https://yoursite.com/pricing and search the output for your price and product names. Vercel found ChatGPT and Claude crawlers fetch JavaScript files but don't execute them. Content that appears only after JavaScript runs may be invisible to them. Googlebot does render.
7. Look for suppression tags. Search the HTML for noindex, nosnippet, max-snippet:0 and data-nosnippet. Google and Bing document these as controls that keep content out of AI answers, and templates often apply them by accident.

Match what you found to the next move.

• Not named, access broken. Fix access first. Until crawlers get a 200 and see your content in the HTML, nothing else on this list can help.
• Not named, access fine. Look at who is cited. When engines cite other sites instead of yours, being named on those sites is associated with getting into the answer. That's an association, not a guaranteed effect.
• Named, facts wrong. Put the correct, specific fact (a price, a number, a one-line description) in plain HTML on a page crawlers can fetch. Then check your own site doesn't state that fact two different ways.
• Named and correct. Re-check in a few weeks. Answers vary between runs.

For Google's surfaces, Google states a page must be indexed and eligible to show with a snippet to appear as a supporting link, with no additional technical requirements. The free AI Overview checker page covers that side in more detail.

A free check is useful, and it has hard limits.

• It's a snapshot. One run on one day is not tracking.
• Your prompts aren't the engine's queries. OpenAI says ChatGPT search typically rewrites a question into one or more targeted queries. Nobody outside the engine sees those.
• Retrieval is discretionary. The same question may be answered from the model's training data or from a live search. Answers without citations may come from a model with a knowledge cutoff months old.
• There's no rank. As of October 2026, Google's AI report in Search Console exposes impressions only, so no position number can be checked against anything.
• Nothing is promised. Fixing access removes blocks. It doesn't make a model pick you.

If you're weighing whether the work pays off, read whether AEO is worth it for a brand and what AI SEO research actually shows.

Steps 1 to 3 are the slow part. The free checker on this site asks up to three buyer questions (your own, or ones written from your homepage), records whether you're named, and scores the answers. It's a snapshot, not continuous tracking.

The AI Visibility Checker on 99Visibility runs the same kind of check. Steps 4 to 7 still matter, because a mention test can't tell you why you're missing. For tools that cover crawler access, robots.txt and other single checks, see free GEO tools and what each kind checks.

Does adding FAQ schema help me appear in AI answers? The evidence says no. Google states structured data isn't required for its generative AI features, and FAQ rich results were deprecated in May 2026. Ahrefs' matched test of 1,885 pages found schema produced no major uplift, with AI Overview citations falling 4.6%. Existing markup does no harm, so keep it if it's accurate. Just don't expect it to get you cited.

Should I allow GPTBot so ChatGPT can find me? GPTBot controls whether OpenAI may use your content for training. It doesn't control ChatGPT search. OAI-SearchBot does that, and OpenAI says each setting is independent. You can block GPTBot and still allow OAI-SearchBot. The common mistake is a blanket AI block in robots.txt that also shuts out OAI-SearchBot, PerplexityBot and Claude-SearchBot.

Why do I get different answers each time I check? Answer engines decide per question whether to search the web or answer from training data, and they rewrite your question into their own queries. Logged-in sessions add memory and custom instructions. Run tests logged out, record the date, and repeat the same questions over time. One run is a sample, not a verdict.

Can my site be allowed in robots.txt and still be blocked? Yes. A firewall, CDN or bot-management rule can return a 403 or a challenge page to an AI crawler that robots.txt allows. OpenAI's own guidance says eligibility needs both the robots.txt permission and a host or CDN that accepts its crawler traffic. Check your CDN's bot settings and logs, not just robots.txt.

Is a free AEO check enough, or do I need paid tracking? A free check answers the first questions: are you named, is the answer right, and can crawlers reach you. If access is broken, fix that before paying for anything. Tracking helps once you need to see change over time across several engines. Whatever you use, ask whether it reports each engine separately rather than one blended score.