ChatGPT shows old pricing for my product: where stale AI answers come from

If ChatGPT shows old pricing for your product, the answer is probably coming from stale public information.

That stale information might be on your own site, an old pricing page, a review site, a marketplace listing, a cached comparison article, or a third-party blog post that still mentions your previous plan names.

The fix is not to ask AI to "update itself." The fix is to make the current pricing easier to retrieve, cite, and trust than the old pricing.

First, find exactly what is stale

Write down the mismatch.

Current price: $49 per month
AI answer: $29 per month
Old plan name: Starter
Current plan name: Growth

Then ask a few direct prompts:

How much does ExampleCo cost?
What are ExampleCo's pricing plans?
Does ExampleCo have a free plan?
Compare ExampleCo pricing to CompetitorCo.

Record which providers show the old answer and whether they cite a source.

Check your own pricing page

Your pricing page should be readable without a full app session.

Run:

curl -L https://example.com/pricing

Look for:

• current plan names
• current prices
• billing period
• free trial or free plan details
• enterprise or custom pricing language
• last updated date, if pricing changes often

If the raw HTML does not contain the current pricing, AI retrieval systems may miss it.

Remove or redirect old pricing URLs

Stale AI answers often come from old pages that still work.

Check for URLs like:

/pricing-old
/pricing-2024
/plans
/compare
/startup-plan
/free-plan

If an old page is still public, decide what it should do:

• update it with current pricing
• redirect it to the current pricing page
• add a clear notice that the page is outdated
• remove it from sitemap and internal links

Do not leave old prices live unless there is a legal or customer-support reason.

Update third-party pricing sources

AI systems often trust third-party pages because they are explanatory and easy to cite.

Check:

• G2
• Capterra
• Product Hunt
• app marketplaces
• affiliate pages
• "best tools" listicles
• old partner pages
• old review posts
• your own guest posts

Search the old price:

"$29" "ExampleCo"
"Starter plan" "ExampleCo"

Then update the pages you control and request corrections where you can.

Add pricing schema when it fits

If your pricing is public and stable enough, structured data can help clarify the offer.

Depending on the page, this may include:

• SoftwareApplication
• Product
• Offer
• AggregateOffer
• FAQPage

Only mark up pricing that is visible on the page. Do not hide schema-only prices that users cannot verify.

Create a pricing FAQ

AI answers often need simple pricing sentences, not only pricing cards.

Add a small FAQ below your pricing table:

How much does ExampleCo cost?
ExampleCo starts at $49 per month for the Growth plan. Annual billing is available, and enterprise pricing is available for larger teams.

Also answer:

• Is there a free plan?
• Is there a free trial?
• What happens when usage increases?
• Is pricing per seat, per workspace, or usage-based?
• Which plan is best for a small team?

These answers are useful for humans and much easier for AI systems to quote accurately.

Put pricing in llms.txt only if it stays current

If your pricing changes frequently, link to the pricing page from llms.txt instead of copying prices into the file.

Good:

- [Pricing](https://example.com/pricing): Current plans, billing options, and free trial details.

Risky:

ExampleCo costs $49 per month.

The second version can become another stale source if nobody updates it.

Re-test with source-aware prompts

After making changes, ask:

What is the current pricing for ExampleCo? Please cite sources.

If the AI still shows the old price, inspect the cited URL. That URL is your next cleanup target.

If it gives no citation, keep strengthening the official pricing page and the pages that link to it.

Scan the pricing page like an AI crawler

Run a NoticeMeAI scan to check whether your pricing page is crawlable, extractable, linked from important files, and clear enough for AI systems to read. Pricing mistakes are usually fixable once the current answer has one obvious source of truth.