ChatGPT Gets Your Software Price Wrong. Here’s What to Do.
A potential customer asks ChatGPT “what is the price for [your product]?” ChatGPT quotes $4,000 per year. The problem is your product’s starting price is $6,000 per year.
The issue is larger than ChatGPT, and points to AI search having a poor track record in accurately quoting prices for technology products and services. In August, VisibilityStack reported that AI got SaaS pricing wrong in some aspect 76 percent of the time.
This error rate is more of the rule than the exception. Part of the reason is how easily AI search repeats what it finds. Recently, researchers concocted a fake concept, made posts about it on sites AI tools like to cite, and found that within a week ChatGPT, Perplexity, Grok, and Google’s AI were all citing it as though it were real.
The takeaway for technology companies is that getting placed on sources that AI typically cites drives AI visibility. If the source gets your pricing model wrong, AI may cite it as published, without running verification checks.
Returning to the ChatGPT error: whether you discover the error while on ChatGPT or during a call with the now frustrated prospect, your first thought will be “how do I fix this?”
What to do first when AI search is wrong about your company
Step one is finding out where the AI search platform, in this case ChatGPT, got its information. There are two places where ChatGPT gets its information, with the second source providing more of a solution:
ChatGPT’s Training Data
ChatGPT’s memory comes from training data, and all answers start here. If the training data is incorrect, there is not much you can do short term. You should focus on getting the right information onto sites that ChatGPT searches.
ChatGPT Searches
ChatGPT runs searches to supplement its memory data, and if the incorrect information appears on one of these sites, there is a clearer path to resolve the issue.
How to tell which one you are dealing with:
1. Open a fresh ChatGPT chat, ideally logged out or in a temporary chat, so memory does not color the answer.
2. Ask the pricing question the way a buyer would. “How much does [company’s product] cost?”
3. Scan for a sources list or inline links. If they appear, ChatGPT used search. Click each link to find the source of wrong information.
4. If there are no sources, ask ChatGPT to "Search the web for current pricing." If the answer corrects itself, the wrong number lives in training data. If it does not, a live source is still feeding it.
5. Run the question three or four times with different wording. AI answers vary, and you want to search for a pattern.
Scenario 1: Training data is giving the wrong price
How it appears: ChatGPT quotes a wrong or former year's price with no sources. When you ask it to search, it gets the price right.
This is the slowest case to correct, but it is not impossible. The old number will disappear as OpenAI uses newer models trained on newer data. In the meantime, you need to make your company’s live web so clear on pricing that search will pick it up short term and the page is included in long-term training. Do this by:
Making your corporate pricing page the clearest source on the web. You should state real numbers in plain HTML text vs. placing pricing inside graphics or tabs that load with JavaScript. If you only use a custom pricing model, give a starting price and consider including a range.
Add a date. Your page should say “pricing current as of [month, year].” AI systems value freshness, and a dated page signals that the price is current.
Check your robots.txt. Make sure you are not blocking OAI-SearchBot (search) or GPTBot (training). Blocking GPTBot keeps your updated page out of future training data.
Publish the real price on third-party sources. Think sites like G2, analyst reports, industry forums, and media outlets.
Scenario 2: The wrong price comes from your own site
How it appears: ChatGPT cites a page on your domain that is old or that you did not realize contained pricing.
The first step for correction is to do a deep dive for all places the pricing may appear. This includes old blog posts, help center articles, FAQ pages, comparison pages, case studies, PDFs, and retired landing pages that are still live. Either update the information on the page or 301 redirect retired pages to your new pricing page.
Next, check your structured data to see if your pages use Product or Offer schema. You want the price in the code to match the price on the page.
You can also ask for a recrawl by submitting updated URLs through Bing Webmaster Tools, and turn on IndexNow, if supported.
Finally, run a retest in a week or two. Use the same buyer questions again and confirm the citation now points to the corrected page.
Scenario 3: The wrong price comes from a third-party site
How it appears: ChatGPT cites a trade publication, an industry blog, a "best tools" roundup, a comparison article, or an analyst report that lists your old price. The course of action is to ask for a correction:
Find the right contact. For the media, look for a corrections email or the editor of that section. For a blog or roundup, contact the author directly. LinkedIn often works better than a generic contact form.
Make it easy for the site to correct. Name the article, quote the incorrect line, give the correct price, and link to your pricing page as proof.
Consider offering extra information. Do you have a new product screenshot or a report that you can share? The person making the change may appreciate the additional detail. If you are reaching out to the media or an analyst firm, you could offer a briefing with your CEO or a senior executive to talk about updates to the product and how it fits into the overall industry.
Scenario 4: The wrong price is in multiple places
How it appears: The same wrong number shows up across review sites, directories, reseller pages, old press coverage, and Reddit threads. ChatGPT sees the wrong price agreeing with itself and treats it as fact.
When sources agree, AI systems get more confident. That works against you here. You need to tip the balance so the correct number is the one most sources repeat.
Triage by control and influence. Start with what you own, then what you can edit, then what you can request. Have a dedicated plan to cover all information sources for correction and keep track of updates. Concurrently, publish a new page that states the correct price clearly: a pricing update post, a refreshed FAQ, or a case study. This will help to reinforce the new pricing.
For all scenarios, or even if your price is listed correctly, add your sitemap to Bing via Bing Webmaster Tools. This allows Bing to index your site, which, in turn, helps ChatGPT and Copilot find your information faster.
Conclusion
AI search has different means of gathering pricing information, and the published information can be wrong. Do frequent searches for incorrect information with a plan to update the data. Also, build a pricing checklist into any product launch and make sure all pages on your site are updated.