Free guide · 2 Oct 2026
Why AI assistants get your SaaS pricing wrong
Ask an assistant "What is [your product] and how much does it cost?" and you may get a confident, specific, wrong number. That matters: a buyer comparing tools may rule you out before visiting your site. The cause is usually something on the web you can find and fix.
How assistants arrive at a price
With web search on, an assistant fetches a handful of pages (your site, directories, review sites, articles) and writes a summary from them, often with citations. Without search, it relies on what the model saw during training, which can be months out of date. Either way, it has no special access to your billing system: it can only repeat what's written somewhere, and it doesn't always tell an example from a real price.
Six common causes
1. Example numbers that look like your price
Product screenshots and mock-ups often contain sample figures: an invoice total, a proposal, a customer's plan. In one of our own checks (Perplexity Sonar API with web search, 1–2 Oct 2026), an assistant gave an example proposal price from a product's homepage mock-up as that product's public price. A directory profile repeated the same figure. Fix: label sample figures clearly as examples ("Example proposal"), or use obviously fictional values, and keep your real price in plain text nearby.
2. Two of your own pages disagree
A FAQ that says "plans start at $15" while the pricing page lists $12, $29 and $59 (made-up numbers) gives the assistant two sources of truth. In another of our checks (same method, 1 Oct 2026), the answer quoted both sets of prices for the same product in one reply. Fix: search your site for every price mention (FAQ, blog posts, docs, changelog, comparison pages) and update or remove the outdated ones.
3. Stale third-party profiles
Directories, marketplaces, review sites and launch listings often keep the price you entered when you first listed. Assistants cite these pages a lot. Fix: list every profile you control, update the price and plan names, and ask the owners of roundup articles to correct outdated figures. Your citations (see the checking method) show which pages matter.
4. The price only exists in JavaScript or an image
If your pricing table is rendered by JavaScript, loaded from an API after page load, or exists only as an image, crawlers that don't run JavaScript may see no price at all. They'll then fill the gap from other pages. Several major AI crawlers don't render JavaScript (see our JavaScript check). Fix: make sure the plan names and prices appear in the HTML the server sends.
5. Currency, region and billing-period ambiguity
"$8" could be per month billed annually, per seat, or in a different currency for some visitors. If the page doesn't say, the summary may not either. Fix: state the unit next to every price ("$8 per user per month, billed annually; $10 billed monthly").
6. Old plan names in your own content
Launch posts and old changelog entries ("our new Pro plan is $29!") stay public and crawlable. Fix: add a short note to old announcements ("Pricing has changed; see current pricing") with a link.
Add machine-readable pricing
Structured data won't override what other sites say, but it removes ambiguity on your own site. A minimal SoftwareApplication with an Offer (schema.org) states the name, category, price and currency explicitly. Google documents the properties it supports for software apps. The values must match what's visible on the page.
{
"@context": "https://schema.org",
"@type": "SoftwareApplication",
"name": "YourProduct",
"applicationCategory": "BusinessApplication",
"offers": {
"@type": "Offer",
"price": "19.00",
"priceCurrency": "USD"
}
}
A 20-minute pricing audit
- Ask 2–3 assistants "What is [product] ([domain]) and how much does it cost?" and save the answers and citations.
- For each wrong figure, open the cited page and find where it came from.
- Fix your own pages first (mock-ups, FAQ, old posts), then the profiles you control, then ask third parties.
- Confirm your pricing page shows plans and prices in the raw HTML.
- Re-ask in 2–4 weeks. Changes on third-party pages can take a while to be picked up.
Want an automated check?
Our AI Visibility Report ($49 PDF) is generated by software. It asks Perplexity and Grok what your product is and what it costs, and puts the quoted prices side by side with the prices found on your homepage, along with category visibility for 10 buyer questions. It doesn't cover ChatGPT, Claude or Gemini yet. Or request a free mini report.
More guides
- How to check whether AI assistants recommend your SaaS
- Can AI crawlers read your homepage? A 5-minute JavaScript check
- How to get your SaaS recommended by AI assistants: a checklist
Sources
- schema.org, Offer: schema.org/Offer; SoftwareApplication: schema.org/SoftwareApplication
- Google Search Central, Software App structured data: developers.google.com
- Vercel, The rise of the AI crawler (JavaScript rendering by AI crawlers): vercel.com
- Examples marked "our own checks" are single API runs we made on 1–2 Oct 2026. They're anonymised and illustrate causes; they aren't statistics.
Independent guide; not affiliated with OpenAI, Anthropic, Google, Perplexity or xAI. Product names are trademarks of their owners.