Why AI Won't Quote Your Prices — and How to Fix It

If you ask ChatGPT, Gemini, or Perplexity what a business charges and it answers “you'll need to contact them for pricing,” that's usually not the model being cautious — it's the model not having a clear, trustworthy price to repeat. AI engines won't quote a price they can't find in plain text, can't tell is current, or can't confirm applies to the right product. If you want AI to state your price, you have to publish it somewhere stable, readable, and unambiguous — or accept that the model will route the buyer to a competitor who did.

I spend most of my time looking at why AI recommends one business over another, and pricing is one of the most overlooked pieces of it. Founders obsess over whether AI mentions them. Far fewer ask whether AI can actually answer the next question a buyer has, which is almost always some version of “what does it cost?” When the model can name you but can't price you, you've won the recommendation and lost the moment right after it.

Why AI hedges on price

Language models are tuned to avoid confidently stating facts they're unsure about, and price is a fact that's unusually easy to get wrong and unusually costly when you do. A model that invents a wrong number creates a bad experience for the user and a support headache for the business. So the engines lean conservative: if they can't find a price they trust, they default to “contact them” rather than guess.

There are a handful of reasons the price you think is on your site isn't a price the model will repeat:

• It's gated behind a form or a “call for a quote.” If the only way to see the number is to fill something out, there's no text for the model to read. As far as AI is concerned, you don't have a published price.
• It lives in an image or a PDF. A price baked into a graphic, a menu image, or a downloadable rate sheet is often invisible or unreliable to a crawler. Multimodal reading is improving, but I wouldn't bet my pricing visibility on a model correctly OCR-ing a decorative menu.
• It's inconsistent across sources. Your site says one thing, an old directory listing says another, a third-party review mentions a number from two years ago. When sources conflict, the model's safest move is to not commit to any of them.
• It has no context. A bare “$99” with no indication of what it buys, who it's for, or whether it's monthly or one-time gives the model nothing safe to say. Ambiguous pricing reads as risky pricing.
• It looks stale. If the page has no sense of being current, the model may decline to quote a number it can't confirm still applies — especially in categories where prices move.

None of these are the model being difficult. Each one is a reasonable response to missing or untrustworthy information. The fix, in every case, is on your side.

What it costs you to be un-quotable

It's tempting to treat “AI won't quote my price” as a minor gap. It isn't, because of what happens in the buyer's head at that exact point. Someone has asked an assistant for a recommendation, gotten your name, and is interested enough to ask what you charge. That's a high-intent moment. When the answer is “I don't have pricing, you'll have to reach out,” you've added friction precisely when the person was closest to acting.

Worse is what the model often does instead. If you're un-quotable but a competitor publishes clear pricing, the model has an easy, helpful answer available — theirs. In a side-by-side, “Their plans start at $149/month, all-inclusive” beats “contact them for a quote” almost every time, because one of those answers actually helps the person and the other sends them to do homework. You can be the better business and still lose the comparison on nothing but price transparency.

And in categories where the model feels pressure to give a number, the real risk is it estimates. An AI guessing your price from category averages or outdated mentions can anchor a prospect to a figure you'd never quote — too low, so they feel misled later, or too high, so they never call. My read is that a wrong number working against you is worse than no number, and you don't get to pick which one the model produces when you leave the gap open.

How to make your pricing quotable

The goal is simple to state: put a clear price, in plain text, in a stable place, with enough context that a model can repeat it without feeling like it's taking a risk. Here's how I'd approach it.

Publish an actual number in actual text. Not an image, not a PDF, not a form. A real pricing section or page with the figures written out as readable copy. If you've been hiding price to force a sales conversation, understand the trade you're making: you may be protecting the call while losing the AI recommendation that would have produced it.

Give every price its context. A number alone is ambiguous; a number with a job is quotable. State what it includes, whether it's one-time or recurring, who it's for, and any obvious conditions. “$2,500 one-time for the standard onboarding, which covers X, Y, and Z” is something a model can confidently hand to a buyer. “Starting at $2,500” is something it has to hedge.

Handle “it depends” honestly with ranges and anchors. Plenty of real businesses can't post one flat price, because the work varies. That's fine — AI can work with structure. Give a starting-at figure, a typical range, or tiered packages with representative prices. “Most projects land between $8,000 and $20,000 depending on scope; our most common package is $12,000” is far more useful, and far more quotable, than “custom pricing.” You're not committing to a quote; you're giving the model an honest anchor.

Be consistent everywhere your price appears. Reconcile your site, your profiles, your directory listings, and anywhere a number might linger. If a model finds the same figure in several trustworthy places and nothing contradicting it, that corroboration is exactly what lets it quote you with confidence. Conflicting numbers are what make it clam up.

Mark it up with structured data. Where it fits, use schema for your offers and price specifications so the price is labeled as a price, in machine-readable form, rather than left for the model to infer from surrounding copy. Schema isn't magic and it won't rescue a number that's wrong or buried, but for a clean, current price it removes ambiguity about what the number means.

Keep it current, and make currency visible. Prices change; stale prices quietly poison trust. When you update, update everywhere, and don't be shy about signaling recency on the page. A pricing section that's obviously maintained is one a model is more willing to repeat.

One more step that costs nothing: actually check what the engines say today. Open ChatGPT, Gemini, and Perplexity and ask what you charge, the way a buyer would. The answer tells you exactly where you stand — whether the model quotes you, hedges, sends people to a competitor, or invents a number. Do this before and after you fix your pages, and you'll see the change for yourself rather than guessing. It's the same thing I tell people about every part of AI visibility: don't assume what the model knows, go ask it, because the gap between what you published and what the model repeats is where all the work is.

A small warning on that check: the first answer you get isn't gospel. Ask a few times, and ask in a couple of phrasings — “how much does X cost,” “what are X's prices,” “is X expensive.” Models vary run to run, and you want the pattern, not a single lucky or unlucky response. If most of your runs still end in “contact them,” you have a pricing-visibility problem no amount of other optimization will fix on its own.

When not to publish a price — and what to do instead

I want to be honest that “publish your prices” isn't universal advice. Some businesses have genuine reasons not to post a flat number: pricing really is bespoke, it's strategically sensitive, or the category norm is negotiated deals where a public figure would do more harm than good. If that's you, don't force a fake number onto a page just to feed the model.

But “we can't post one price” is not the same as “we should say nothing.” Even here, you can give AI something to work with: the factors that drive your pricing, a transparent explanation of how you quote, a typical range for a representative engagement, or what a buyer should expect to budget. That lets the model set accurate expectations and still point people to you, rather than leaving a vacuum that a competitor's clear pricing fills. The worst outcome isn't a published price you'd rather not share — it's a buyer at peak intent getting handed to someone else because your number wasn't there to repeat.

The through-line across everything here: AI visibility doesn't end at getting mentioned. It extends to every follow-up question a buyer asks, and price is the one they ask most. Make yours findable, readable, contextual, and current, and you turn “contact them for pricing” into an answer that actually moves someone toward you.

About the author

Scott Tischler is the Founder & Chairman of AIrecommend.ai and a practitioner-authority on AI search and Answer Engine Optimization. With 20+ years in marketing technology — including American Express, MetLife, and UBS — and executive and professional study at Wharton, Harvard, and Oxford, he helps businesses become the ones AI recommends.