The price is right on your site. Nobody's reading it there
You ran a promo. Dropped the 32oz bottle from $34 to $29, pushed it live, confirmed the number on your own product page. An hour later a shopper asks ChatGPT what it costs and gets $34, sourced from your Amazon listing that hasn't synced yet.
This isn't a bug in your CMS. It's a decision the engine made about who to trust. When a product is listed in two places, an AI system has to pick a source of truth for the price, and it doesn't automatically pick yours. It picks whichever listing looks most authoritative and easiest to parse. Large marketplaces usually win that contest because they carry clean structured data, high crawl frequency, and a mountain of corroborating signals like reviews and Q&A.
So the place you control least becomes the place that answers for you. And the price you just changed is the one nobody quotes.
Why the marketplace copy beats your own page
Three things stack against your site. First, structured data. Marketplace listings ship consistent, machine-readable price and availability markup at scale. If your product page uses thin or inconsistent schema, the engine has a cleaner alternative and takes it.
Second, authority signals. An assistant weighing two prices for the same SKU leans toward the source with more traffic, more links, and more transactional history. That's almost always the marketplace, not your DTC page.
Third, update lag works backwards from what you'd expect. You assume the freshest source wins. But the engine may have cached the marketplace listing more recently than it recrawled your site, so a six-day-old third-party number can outrank a price you changed this morning. The result is predictable: you eat the margin on the lower price while the AI advertises the higher one, or you lose the sale to a competitor the assistant surfaces alongside the stale quote.
Owned product page vs marketplace listing, through the engine's eyes
| Comparison category | What the engine checks | Your product page | Amazon / marketplace listing |
|---|---|---|---|
| Price schema quality | How cleanly the price is marked up | Often partial or missing | Complete, consistent markup |
| Recrawl frequency | How recently the engine refreshed it | Days to weeks | Hours to days |
| Authority signals | Links, traffic, and corroboration | Brand site only | Reviews, Q&A, sales rank |
| Who controls the price shown | Who owns the number the AI reads | You | The seller feed, which lags |
| Default source of truth | Who wins when prices conflict | Loses the tie | Wins the tie |
Find out which source is actually answering for you
You can't fix a pricing conflict you can't see. The first move is to ask the engines directly and read the citation, not just the number. If the price is wrong, check which URL it came from. If it came from a marketplace, you've found the leak.
Crescive does this continuously. It runs pricing questions across ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google AI Overviews, records the exact price each one quotes, and tags the source it cited. When an engine is pulling from a stale marketplace listing instead of your current page, Crescive flags it and shows the gap between what you charge and what the AI says you charge. Then it drafts the fixes, tighter price schema, clearer availability signals, the kind of structured data that gives your own page a fair shot at the tie, and routes them through a human approval gate before anything ships. After the change, it re-checks and shows whether the citation flipped back to you.
A pricing-source audit you can run this week
- Pick your most recently repriced SKUs and ask each AI assistant what they cost, then record both the number and the cited source URL.
- Flag any answer that quotes a marketplace listing or an out-of-date price.
- Compare your product page schema against the marketplace listing's markup for price, currency, and availability, and close the gaps.
- Confirm your product pages are being recrawled near the cadence of the marketplace copy, and submit updated pages after any price change.
- Re-ask the engines a few days later and verify the citation moved to your page and the price matches.
Key takeaways
- When a SKU lives on your site and a marketplace, AI assistants often default to the marketplace as the source of truth for price, so your freshest number loses.
- Marketplace listings win because of cleaner structured data, higher recrawl frequency, and stronger authority signals, not because they're more accurate.
- Track which source each engine cites for pricing, not just the number, and strengthen your own page's schema so it can win the tie. Crescive monitors this and flags when the wrong source is winning.
FAQ
Why does an AI quote my Amazon price instead of the price on my own website?
Because the AI picks a single source of truth for the price, and marketplace listings usually look more authoritative. They carry complete, machine-readable price schema, get recrawled more often, and have supporting signals like reviews and sales history. If your product page has thinner structured data or was recrawled less recently, the engine defaults to the marketplace copy, even when that price is older than the one you just updated on your site.
How do I know which source an AI is using for my product's price?
Ask each assistant for the price and read the citation it provides, then check the source URL against the quoted number. If the price is stale, trace it to its source. Crescive automates this across ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google AI Overviews, recording the price each engine quotes and the source it cited, and flagging any time a stale marketplace listing is winning over your current product page.