What to Do When ChatGPT, Gemini, or Perplexity Shows Wrong Information About Your Business

Wrong business information in ChatGPT, Gemini, or Perplexity can be more damaging than a low search ranking. A chatbot may describe an old service, list the wrong opening hours, confuse your company with another business, or claim that you do not serve a location you have covered for years.

These systems do not maintain a single official database for every company. Depending on the product, an answer may come from a live web search, an indexed page, a business directory, a review site, a data provider, or model knowledge that is no longer fresh.

Small businesses are especially exposed because their information is often scattered. Your Google Business Profile may show one phone number, an old marketplace listing may show another, and a social profile may still mention a service you stopped offering in 2023.

The result is a messy signal environment. An AI system sees several versions of your business and has to guess which one is reliable. That guess can sound confident even when the underlying evidence is weak.

A useful first distinction is between a factual error and an incomplete answer. If an AI says your bakery is closed on Sundays when it is open from 9 a.m. to 2 p.m., that is a direct factual error. If it fails to mention your gluten-free menu, the answer may simply reflect missing content rather than a contradiction.

You should also separate a one-off response from a repeatable pattern. Test the same question in a new conversation, use slightly different wording, and check more than one platform. Record the date, exact prompt, answer, and any cited sources before you start making changes.

This evidence matters because AI answers can vary. A screenshot that does not include the question, date, and source links is difficult to investigate later. Treat each result like a customer support ticket, not like a personal insult from a robot having a bad morning.

No. Publishing one correction page cannot force ChatGPT, Gemini, or Perplexity to change a response. These systems may crawl, retrieve, index, or use pages on different schedules, and some answers are generated from sources you cannot directly control.

A correction page still matters because it creates a clear reference point. Instead of making a chatbot reconcile five scattered descriptions, you give it one public page that states the fact plainly, identifies the business, includes an effective date, and links to supporting information.

Think of this page as a canonical information record, not a magic reset button. It should answer questions such as what you offer, where you operate, how customers contact you, when the information was last verified, and which details have changed.

Use direct language. “Bright Dental accepts new patients at its Austin location Monday through Friday from 8 a.m. to 5 p.m. The practice does not offer emergency care” is easier to interpret than “We are here for all your dental needs.” Specificity wins because it leaves less room for guesswork.

The page should be publicly accessible, indexable, and linked from relevant parts of your site or hosted blog. Add it to your sitemap when appropriate, then request inspection in Google Search Console’s URL Inspection tool so you can confirm whether Google can access the page.

Structured data can help search systems understand entities and properties, but it does not guarantee an AI citation. Google explains the role and limits of structured data in Search, and the same principle applies here: markup supports clarity, while accurate visible content does the heavy lifting.

For sensitive claims, include evidence rather than marketing language. A clinic can link to its appointment page, a restaurant can link to its current menu, and a SaaS company can link to documentation that confirms a feature. The correction should be useful to a customer even if no AI ever quotes it.

A good claim page is short enough to scan but complete enough to remove ambiguity. Start with the official business name, the legal or public-facing identity, the primary location, the service area, and the best customer contact method.

Next, create a fact table with fields that frequently appear in AI answers. Include current hours, holiday exceptions, products or services, starting prices if you publish them, accepted payment methods, booking requirements, delivery areas, return policies, and the date each detail was last checked.

Add a section called “What has changed” when the error involves outdated information. For example: “As of September 1, 2026, Northside Pilates no longer operates at 14 Oak Street. Classes are now held at 88 King Avenue, and the phone number is unchanged.” This gives retrieval systems and humans a clear timeline.

Use one fact per sentence whenever possible. Avoid burying the correction in a long brand story, because a customer asking about your hours should not need to excavate your origin story like an archaeologist with Wi-Fi.

Include supporting links that are stable and owned by you. A booking page, current menu, pricing page, product documentation, or verified profile is more useful than a vague statement that “many customers know this.”

For a local business, use the same spelling for your name and address across your pages. For an online store, state shipping regions, inventory conditions, and customer support channels. For a SaaS company, clarify the product category, core use case, integrations, pricing model, and who the product is not designed for.

You can use headline and lead sentence formulas for AI answer engines to make important facts easier to find, but do not force awkward keywords into the page. Clear answers beat clever copy when the goal is correcting a business record.

If the error affects several customer questions, create a small information hub rather than one enormous page. A central official information page can link to focused pages for hours, locations, services, pricing, and policies. That structure gives each important fact a clear home and makes future updates less risky.

There is no universal correction time. A live-search answer may reflect a newly published page quickly, while a response based on a model’s stored knowledge may not change until the system receives newer evidence or is updated. Perplexity, Gemini, and ChatGPT can also use different retrieval sources for similar questions.

A practical expectation is to measure in stages. Owned-page discovery may take days or longer, external directory updates may take several days to a few weeks, and stable behavior across repeated AI prompts can take longer. Treat these as operating estimates, not promises from any platform.

The first common mistake is arguing with the chatbot instead of repairing the source ecosystem. Telling an AI “that is wrong” may improve one conversation, but it does not update your Google Business Profile, directory listing, website, or the information used in future sessions.

Another mistake is publishing ten slightly different correction pages. This creates competing URLs and increases the chance that one page says “open Sundays” while another says “closed Sundays.” Pick a canonical page, link to it consistently, and update it when the fact changes.

Do not use fake reviews, keyword stuffing, copied directory descriptions, or aggressive requests for favorable answers. Brand defense is about making accurate information easier to verify, not manufacturing popularity signals.

Avoid changing facts without an audit trail. If your hours change for a holiday, record the start and end date. If a product is temporarily unavailable, say that clearly instead of deleting the page and leaving customers to find an outdated version elsewhere.

Use Google’s official guidance on creating helpful, reliable content as a quality check. The page should serve a real customer first, explain the business plainly, and demonstrate who is responsible for keeping the information current.

Finally, monitor impact beyond citations. Track calls, bookings, contact form submissions, branded searches, and customer questions about the corrected fact. A chatbot answer that changes is encouraging, but the real victory is fewer confused customers and fewer conversations that begin with, “I saw online that you no longer do that.”