Blog Playbooks

When AI assistants get your prices or stock wrong

Where assistants get product facts, how to trace a wrong price or out-of-stock claim to its source, fix it, and confirm the fix with repeated sampling.

A wrong price or an "out of stock" claim is worth tracing to a source before you try to change anything. The platforms describe where price facts come from, including third-party providers, the first merchant listed and update delays. A stale copy elsewhere is one possible cause, but Dopbot did not measure causes. So the job is not editing the answer: it is finding which source the answer read and correcting that.

Where do assistants get prices and stock status?

As of October 2026, the platforms describe it in general terms:

  • OpenAI's shopping help says the prices shown with products are supplied by third-party providers, alongside structured metadata such as price and description from first-party and third-party sources. It adds that a price or shipping change by a merchant may take some time to appear. The price in the first response typically reflects the first merchant listed, which may not be the lowest. For Shopify merchants, product data reaches ChatGPT through Shopify Catalog.
  • OpenAI's web search help warns that search results and citations can be incomplete, outdated or incorrect.
  • Perplexity says it is more likely to recommend merchants that provide deeper product details such as availability, reviews and pricing. For questions about availability, shipping or prices, it sends people to the brand or retailer.
  • Google Merchant Center lists the delay between a change on your website and the matching change in your product data among the usual reasons for wrong availability.

Our October 2026 sample shows which kinds of pages appear in each engine's source lists. Percentages are the share of that engine's answers whose list holds at least one domain of the type:

Domain type in the source listOpenAI's API with web search (gpt-6-luna), 800 answers, pages linked in the answerPerplexity's Agent API with the Sonar model (perplexity/sonar), 787 answers, pages its search retrieved
Retailer (hand-made list)17.6%76.7%
Own-site match46.6%62.0%
Forum or video1.3%3.3%

Read this as a map of where facts can come from, not as a ranking of influence: a domain in a list does not show that it shaped an answer. Counts cover all 160 categories in the US, Canada and Mexico and all five questions of the October 2026 edition, including the alternative-to-a-brand question. The retailer list has 49 domains chosen by hand and misses some stores, so retailer shares are lower bounds; forum or video means reddit.com, youtube.com, quora.com, tiktok.com, instagram.com and facebook.com. An own-site match means the list holds a domain named after a recommended brand, such as samsung.com for Samsung; it misses country domains, parent companies and other store names.

Dopbot's index does not measure whether the prices or stock statuses in answers were correct, so we cannot tell you how often assistants get them wrong. The table also says nothing about where any price in an answer came from: the index did not record prices, and it samples developer APIs, not the ChatGPT app. Use it only to see that retailer domains from our list and brand websites are both common in these source lists.

Which kind of error is it?

This table is Dopbot's reading of the platforms' documentation and common feed practice, not a measured diagnosis: treat each row as a place to look, and confirm it with the sources the answer shows.

What you seePossible originWhere to check
An old sale priceA stale feed, stale markup, or a retailer still showing the saleYour page's Product markup, Merchant Center diagnostics, the retailer listing
"Out of stock" for an item you haveAvailability in markup or a feed that updated late; markup built by JavaScriptThe source URLs the answer shows; Google says dynamically generated markup can make Shopping crawls less frequent and less reliable for price and availability
A price that is right on your site but not in the answerAnother merchant's price shown firstThe other merchants ChatGPT lists for the product, if it lists any
A feed price that differs from your pageA mismatch between product data and websiteThe Needs attention tab in Merchant Center
Wrong shipping terms or datesShipping terms lag, which OpenAI says can happenYour shipping page, your feed, the retailer's terms
A wrong fact quoted from a review or list pageA third-party pageThat site's editor

How do you fix it?

  1. Capture the evidence. Save the exact prompt, the assistant, the date and time, your location setting, any source URLs shown, and a screenshot. OpenAI suggests asking again with a specific source, date or location when a result looks wrong.
  2. Open every source the answer shows and compare it with your product fact sheet. A stale source may be a page you do not control.
  3. Fix your own page. Show the price and availability in visible text and in Product offer markup. Google's merchant listing guide requires the active price, with a priceCurrency, recommends Product markup in the initial HTML, and suggests validating with the Rich Results Test. The structured data guide covers Shopify specifics.
  4. Fix your feeds. Schedule Merchant Center updates to coincide with store updates. Google's automatic item updates are meant for sporadic problems and are not a replacement for regular product data updates. If you sell on Shopify, check the data Shopify Catalog holds. OpenAI's feed specification asks merchants to update availability when an item sells out and to update the feed when a sale starts or ends; as of October 2026 its developer documentation says feed onboarding is available to approved partners. The feeds post lists the fields.
  5. Fix retailer and marketplace listings, and ask the retailer to correct what you cannot edit.
  6. Use the feedback route where it exists. As of October 2026, OpenAI's shopping help invites a thumbs-up or thumbs-down on the product, a report of products with incorrect information, or its content reporting form. Perplexity's help center says to use the flag icon below an answer to report issues, or to open a support ticket, and to include the query URL and a description of the error. Its Instant Buy page sends product-specific questions to the brand or retailer, which is another reason to fix the source.
  7. Write it down. A short log of the date, the source fixed and the date you re-asked turns a hunch into evidence.

How do you confirm the fix?

One good answer proves little, because assistants return different answers to the same question. Re-ask the same prompt many times, in fresh sessions, from the same location, spread over several days, and count the answers that state the wrong fact.

The 95% Wilson intervals below use illustrative counts, not data. They show how many answers it takes to say a wrong fact has become rare, not to prove it is gone. The intervals assume each question is an independent draw, so use fresh sessions.

Wrong answers out of answers askedObserved ratePlausible true rate
6 of 20, before the fix30.0%14.5% to 51.9%
0 of 20, after the fix0.0%0.0% to 16.1%
0 of 40, after the fix0.0%0.0% to 8.8%
0 of 100, after the fix0.0%0.0% to 3.7%

With 20 answers, a clean result is encouraging but still allows an error rate up to 16.1%. Keep sampling until the upper end is a rate you can live with, and see the post on how many answers to trust for the arithmetic. Wait before judging. None of the OpenAI or Perplexity pages we checked gives a time for a corrected price to show up in an answer, and OpenAI says it is working on faster methods. Google says a website check after an availability fix in Merchant Center can take up to 12 hours. Shopify gives a time for removal: up to 7 days after you switch off Shopify Catalog access for third-party channels.

What can you not control?

  • Labels and summaries. OpenAI says labels such as "Budget-friendly" are generated by ChatGPT and are not guarantees, and that it may simplify product titles and write review summaries from public websites.
  • Other sellers. If several merchants sell your product, the price shown first may not be yours.
  • API versus app. Our index samples developer APIs, and the apps can differ in location, personalization and product interfaces. Check the app yourself, and read API vs app answers.
  • Timing. A correct source does not change an answer that has already been given to someone.

Frequently asked questions

Can I use the same answer count if I check the ChatGPT app by hand?

The counting works the same way: ask the same question repeatedly in fresh sessions and count the wrong answers. App answers can differ from API answers in location, personalization and product interfaces, so record your location and settings with every answer. The API vs app post explains the differences.

Do I need a product feed for ChatGPT?

As of October 2026, OpenAI says no, but that feeds give you greater control over how your products appear, and it says Shopify and Etsy catalogs are already integrated with no extra setup. Its developer documentation says feed onboarding is available to approved partners. A feed is one way to keep price and availability current; it does not by itself prevent a wrong price.

How we write: drafted with AI tools from Dopbot Index data and public sources; every number and outside fact is traced to its source in a claims record and checked before publishing. More about how we write; corrections to hello@dopbot.com.