What's an affordable way for a small business to monitor what AI chatbots say?

Most small businesses can monitor what AI chatbots say for close to nothing. The routine has three layers:

• Free checks confirm AI crawlers can reach your site and give you a snapshot of how engines answer buyer questions.
• A monthly manual protocol repeats the same questions on the same engines and logs every wrong fact.
• A paid plan takes over once the manual work stops fitting in your calendar.

The reasoning is simple. The errors that hurt a small business are specific facts: a price, opening hours, a location, a service you stopped offering. You can see those by asking. What you can't do by hand is repeat that across many questions, engines and brands every month. That repetition is what a tool automates.

Keep one limit in mind. Monitoring, paid or free, tells you what's wrong. It doesn't control what the model says. The fixing happens on your own pages and on the sites the engines cite.

Do these once, then again after any site rebuild or hosting change.

• Get a snapshot. The free checker on this site shows whether AI engines name you for up to three buyer questions. It's a snapshot, not tracking. For a wider pass, see free AI brand monitoring across several engines.
• Read your robots.txt per bot. OpenAI's crawler page says GPTBot governs training and OAI-SearchBot controls ChatGPT search, and each setting is independent. A blanket AI block can quietly remove you from ChatGPT search.
• Test the firewall. OpenAI's help page says eligibility also requires your host or CDN to allow its search crawler's traffic. A 403 from your firewall blocks the bot even when robots.txt allows it.
• Put key facts in the raw HTML. Vercel found that ChatGPT and Claude crawlers fetch JavaScript files but don't execute them. A price that only appears after JavaScript runs may be invisible to them.
• Check Google eligibility. Google says a page must be indexed and eligible for a snippet to appear as a supporting link in AI Overviews or AI Mode. Look for stray noindex or nosnippet tags.

Block an hour on the same day each month and follow the same steps.

1. Write 10 to 15 buyer questions. Say you run a dental clinic in Austin: "best emergency dentist in Austin", "how much does [clinic] charge for a cleaning", "is [clinic] open on Saturdays". Keep the list fixed.
2. Pick the engines your buyers use. Each one behaves differently, so check them one at a time. Our page on tracking Claude's answers about your brand covers one example.
3. Run them logged out, or with memory off. OpenAI says ChatGPT may use saved memories when rewriting a search query, which makes logged-in results hard to repeat.
4. Log each answer. Record engine, date, whether sources were cited, each fact stated, and whether it was right, wrong or missing.
5. Classify each error. An invented fact and an outdated one need different fixes. See hallucination vs outdated answers.
6. Fix the source. State the correct fact plainly on a page crawlers can fetch. Make it match across your pricing page, about page and footer. If the answer cited a directory or review site, correct your listing there.
7. Re-run next month. The same questions make the comparison fair.

For a ChatGPT-specific version, read monitoring what ChatGPT says about your brand.

A spreadsheet is honest, but it has blind spots.

• Search may be off. Engines decide per query whether to search the web. An answer with no citations may come from training data frozen at a cutoff. As of August 2026, OpenAI listed a February 16, 2026 cutoff for its GPT-5.6 models and August 31, 2025 for the Instant path ChatGPT uses.
• Your prompt isn't the search. OpenAI says ChatGPT search rewrites your question into one or more targeted queries. Nobody outside sees those queries.
• Engines differ sharply. In Otterly's 2026 experiment, Microsoft Copilot produced 2,718 of the 3,934 citations it tracked for one site, while ChatGPT and Google's AI features almost never cited it. A single blended number would have hidden that.
• It isn't live. A monthly check catches errors weeks late. See whether any tool detects misinformation in real time.

Manual checking works until one of these is true:

• Your question list has grown past what you can run in one sitting.
• You manage several brands or locations.
• You need a monthly report for a client, partner or board.
• You want wrong facts flagged automatically against your own verified facts.

As one reference point, 99Visibility publishes these plans: Free at $0 (5 keywords, 1 audit a month, 1 brand, 2 platforms), Starter at $49/mo (25 keywords, 15 audits a month), Growth at $99/mo (50 keywords, 50 audits a month, 3 brands) and Pro at $149/mo (100 keywords, 150 audits a month, 5 brands). Paid plans audit ChatGPT, Google AI Overviews, Gemini, Perplexity and Claude. Details are on the pricing page. For the wider market, see hallucination tracking under $500 a month.

Whatever you buy, skip features with no evidence behind them. Google says E-E-A-T isn't a specific ranking factor, so there's no score to track. Google's AI report exposes impressions only, so an "AI rank" can't be checked. Our checklist for choosing a brand accuracy platform covers the rest.

How often should a small business check AI answers about its brand? Monthly suits most small businesses. Check sooner after a price change, a new location, a rebrand or a site migration, because those are the moments facts drift. Also re-check crawler access after any hosting or CDN change, since a firewall setting can block AI crawlers without anyone editing robots.txt.

Will adding schema markup or an llms.txt file fix wrong AI answers? The evidence says no. Google states structured data isn't required for its generative AI features and that Google Search ignores llms.txt. Ahrefs' matched test of 1,885 pages found AI Overview citations fell 4.6% after schema was added. Existing markup does no harm, but it isn't a lever for AI answers. Stating the correct fact plainly on a fetchable page is the better move.

Why does ChatGPT give a different answer each time I ask? Several things vary. ChatGPT decides per question whether to search the web, rewrites your question into its own search queries, and may use saved memories when logged in. Different models can also serve different users. Run checks logged out, note whether sources were cited, and compare like with like across months.

Can I make an AI chatbot stop saying something false about my business? Not directly. No one outside the AI company controls what the model says. You can correct the pages it reads: your own site and the third-party sites it cites. Make sure crawlers can reach those pages and that the right fact appears in the HTML. Then recheck. Answers from training data change only when the model is retrained.