Fixing AI Answers That Get Your Pricing Wrong

An AI answer about your pricing passes through four stages. The model may answer from trained knowledge (parametric). If it searches, a crawler must be able to reach your page (access), the engine must pull it into context for that question (retrieval), and the model must then use it (generation). A wrong price can come from a failure at any stage.

The common causes, in rough order of how often they are worth checking:

• The price is rendered by JavaScript. Vercel's crawler study found that ChatGPT and Claude crawlers fetch JavaScript files but do not execute them, so they cannot read client-side rendered content. Googlebot does render, so Google surfaces behave differently from the rest.
• Your own pages disagree. The pricing page says $49, an old launch post says $39, a comparison page still lists last year's tiers. An engine that retrieves the wrong page quotes the wrong number.
• Third-party listings are stale. Directories, marketplaces and review sites often carry the price from when you signed up. When an engine cites those sites, it repeats them.
• The answer came from trained knowledge. OpenAI's model docs list a February 16, 2026 knowledge cutoff for the GPT-5.6 family and August 31, 2025 for the Instant path used in ChatGPT. A price you changed after the cutoff cannot appear in an answer given without search.

Access blocks sit underneath all of these. A robots.txt rule, a firewall, or a stray nosnippet tag can keep your correct price out of the answer even when the page is perfect. For the wider version of this problem, see the fix plan for AI getting your brand wrong.

Work through these in order. Each step removes a reason the correct price cannot reach the answer.

1. Record what is wrong, per engine. Ask each platform two or three questions a buyer would ask ("How much does [brand] cost?", "Does [brand] have a free plan?"). Write down the price stated, whether the answer showed citations, and which URLs it cited. The free checker on this site runs up to three buyer questions and shows whether the answers get you right. No citations usually means the answer came from trained knowledge.
2. Confirm the price is in the raw HTML. Open your pricing page with JavaScript disabled, or view the page source. If the plan names, prices, currency and billing period are missing, move them into the server-rendered HTML. Our guide to making SaaS pricing pages readable to AI covers the patterns that break most often, such as monthly and annual toggles that only render on click.
3. Make every page you own say the same thing. Search your site for old prices: blog posts, launch announcements, help docs, comparison pages, PDFs and archived landing pages. Update or remove each one. A check for contradicting facts on your own site helps you find them.
4. Check that each search crawler can reach the page. robots.txt rules are per bot. OpenAI states that each of its crawler settings is independent, and that OAI-SearchBot controls ChatGPT search while GPTBot controls training only. Then confirm your firewall or CDN does not return a 403 to those crawlers, which robots.txt cannot show you. 99Visibility's free AI Crawler Checker requests a page as each AI crawler and shows which ones get turned away.
5. Remove accidental suppression. Look for noindex, nosnippet, max-snippet:0 or data-nosnippet on the pricing page or its template. Google and Bing document these as controls that keep content out of AI answers.
6. Correct the third-party copies. Start with the URLs the engines actually cited in step 1. Update your listing, or contact the site owner if you cannot edit it. Perplexity shows its sources plainly, which makes correcting the sources Perplexity cites a good place to start.
7. Recheck after the engines have had time to recrawl. Rerun the same questions and compare against step 1.

A minimal robots.txt that lets the main search crawlers in looks like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

Structured data is not a fix here. Google says structured data is not required for generative AI search, and an Ahrefs matched test found no meaningful lift in AI citations after pages added it. Existing markup does no harm, but your time is better spent on the steps above.

Run this list against your pricing page and every page that mentions a price.

• Each plan name and price appears in the HTML before JavaScript runs.
• Each price states its currency and billing period in the same sentence ("$49 per month, billed monthly").
• Annual and monthly prices are both visible without a click or toggle.
• Free plan or trial terms are written out in plain text, not only shown as a badge.
• No other page on your domain states a different current price.
• Old prices on archived posts are updated or clearly marked as historical.
• robots.txt allows OAI-SearchBot, PerplexityBot and Claude-SearchBot if you want to appear in those engines.
• Your firewall or CDN returns a 200, not a 403 or a challenge page, to those crawlers.
• The pricing page has no noindex, nosnippet or max-snippet:0, and no data-nosnippet around the price.
• The page is indexed in Google and in Bing (check Search Console and Bing Webmaster Tools separately).
• Directory, marketplace and review listings show the current price.

If you are weighing whether to show prices at all, read whether to publish prices for AI answers. An engine that finds no price on your site has to take one from somewhere else.

Judge results per engine and per question, not as one blended number.

Good results

• Search-backed answers state your current price and cite your pricing page.
• Different engines agree on the price, even if they word it differently.
• When an answer cites a third-party site, that site now shows the correct figure.
• Answers without citations are wrong only in ways that match the model's knowledge cutoff, and search-backed answers on the same engine are right.

Bad results

• A search-backed answer cites your own site and still gets the price wrong. Look for a second page on your domain with an old number.
• Answers cite a directory or review site with last year's price. That listing is your next fix.
• An engine never cites your site at all. Check crawler access and indexing for that engine.
• Google surfaces are right but ChatGPT, Claude and Perplexity are wrong. That pattern fits JavaScript rendering, because Googlebot renders and the others do not.
• Answers mix your price with a competitor's. Check whether comparison pages, yours or others, present the two side by side ambiguously.

Each engine has its own quirks. See ChatGPT's trained knowledge versus search and fixing wrong facts in Microsoft Copilot, which grounds on Bing's index.

Pricing changes more often than most brand facts, so check on a schedule.

• Every price change. Update the pricing page, then search your own site and your listings the same day. Rerun the buyer questions a week or two later.
• Monthly, otherwise. Rerun the same questions on each engine and compare against your last record. Keep the questions fixed so the results are comparable.
• After any infrastructure change. A new CDN rule, firewall setting or site rebuild can block crawlers or move prices into JavaScript without anyone noticing. Cloudflare changed its defaults on September 15, 2026 to block Training and Agent crawler categories on ad-displaying pages for new domains and free-tier customers, while leaving Search allowed. Default changes like that can alter access without you editing anything.
• When a new model ships. A newer knowledge cutoff can fix stale trained-knowledge answers, or introduce a different stale price.

Be clear with your team about the limits. Retrieval is per query and discretionary on every platform: ChatGPT searches at the model's discretion, Claude.ai search is a user toggle, and Google AI Overviews often do not trigger. So the same question can get a search-backed answer one day and a trained-knowledge answer the next.

Google says eligibility for AI Overviews and AI Mode requires a page to be indexed and eligible for a snippet, and that indexing and serving are not guaranteed. No engine publishes how it chooses between your pricing page and a third-party listing.

What you control is whether the correct price exists, is consistent, and can be reached. That raises the odds of an accurate answer. Correcting a model inside a single chat is a different matter, covered in whether correcting ChatGPT in a chat fixes it for everyone.

Should I block AI crawlers from my pricing page to stop wrong prices? Blocking usually makes it worse. If search crawlers cannot fetch your pricing page, engines have to take a price from somewhere else, often an old listing. Blocking GPTBot only affects training, not ChatGPT search, which uses OAI-SearchBot. The better move is to keep search crawlers allowed and make sure the page states one clear, current price in the raw HTML.

Why does Google's AI get my price right but ChatGPT doesn't? A common reason is JavaScript. Googlebot renders pages, so it sees prices that load client-side. Vercel's study found ChatGPT and Claude crawlers fetch JavaScript files but do not execute them. If your price only appears after scripts run, Google surfaces may show it while other engines miss it. View your page source to confirm.

How long does it take for AI answers to reflect a new price? Nobody publishes a timeline. Search-backed answers depend on when each engine recrawls your page, which varies by platform. Answers from trained knowledge do not change until a newer model with a later cutoff ships. Recheck a week or two after a change, then monthly, and record whether each answer showed citations.

Will adding schema markup fix wrong prices in AI answers? There is no evidence it will. Google says structured data is not required for its generative AI features, and an Ahrefs matched-control test found AI citations barely moved after pages added schema. Existing markup does no harm. A plainly written price, in server-rendered HTML, on a page crawlers can reach, is the fix with the stronger evidence.

What if the wrong price comes from a review site I don't control? Start by updating any listing you can edit yourself, since many directories and marketplaces let vendors change their own profile. For sites you cannot edit, contact the owner with the correct figure and a link to your pricing page. Focus on the URLs engines actually cite in their answers, because those are the copies that reach buyers.