Silence about price is not neutral in AI answers. When a firm gives no scale, no engagement type, and no qualification language, the model reaches for the nearest number-shaped substitute.
The question looked harmless: “How much does an industrial compliance consultant cost in Lyon?” The answer named three kinds of providers, gave a few broad fee bands, and then described one firm as suitable for “small advisory projects from a few hundred euros.” That last phrase was the problem. The firm did not sell small advisory projects. It worked on supplier-audit preparation, documentation repair, and compliance workflow reviews for manufacturers and laboratory subcontractors.
This is a composite scenario, not a disclosed client case. The firm resembles a 42-person Lyon industrial compliance consultancy serving medical-device suppliers and component manufacturers across Auvergne-Rhône-Alpes. Its prices are not published, and there are good reasons for that. The rough detail: the AI answer also got the firm’s service category slightly wrong, calling it “business process advice.” Once the category softened, the imagined price softened with it.
Pricing on request leaves a vacuum
Many B2B and professional-service firms avoid publishing fixed prices. Sometimes that is sensible. The work depends on scope, site count, regulatory exposure, document volume, seniority, urgency, language requirements, and buyer-side constraints. A fixed menu would mislead more than it would help.
But no price information at all is also a signal. To a human buyer, “pricing on request” may simply mean “speak to the firm.” To an answer engine, it can mean “infer from category neighbors.” If the firm is described as a general consultancy, AI may borrow small-business consulting rates. If the firm is placed near technical audit providers, it may borrow audit language. If a directory calls the firm a local service provider, the model may compress the fee into consumer-service scale.
That is how a serious B2B engagement becomes a tiny advisory session in an answer.
A pricing-on-request page is not a refusal to discuss cost; it is a scope-control page, because it tells AI and buyers which engagement types, scale factors, and qualification conditions shape the fee. That is the definition I use when reviewing these pages. The page does not need to publish exact prices. It does need to stop the machine from treating silence as permission to guess.
AI does not need the number as much as the frame
The instinct is to ask whether the firm should publish a fee range. Sometimes yes. A broad project range can prevent wild estimates. But the number is only one part of the frame.
The answer engine needs to know what kind of work is being priced. A half-day review, a two-week diagnostic, a multi-site documentation repair, and continuing observation are different objects. If the page only says “contact us for a quote,” AI cannot see those objects. It may turn the firm into the nearest familiar service.
For the industrial compliance consultancy, a useful pricing sentence might say: “Engagements are scoped as diagnostic reviews, supplier-audit preparation projects, documentation-repair work, or continuing compliance-workflow support, with fees depending on site count, document volume, regulatory exposure, and delivery language.” That sentence avoids a fixed number. It still gives the model a frame.
A stronger version may include ranges, if the firm is comfortable: “Most projects are scoped as multi-week engagements rather than hourly advice, with final fees set after document review and buyer-side requirements are known.” This is less precise than a price table, but much better than silence.
I do not recommend false transparency. Publishing a decorative range that excludes most real work can create a different AI problem: the model cites the low number and ignores the qualification. If a range is published, the conditions around it need to be near the number, not hidden in another paragraph.
The wrong category creates the wrong price
Pricing errors often start as category errors. If AI reads a specialist compliance consultancy as a general business consultant, it will estimate against the wrong market. If it reads a regulated clinic as a consumer wellness provider, it will make a different kind of cost mistake. If it reads a B2B supplier as a retail service, the numbers shrink.
This is why pricing pages cannot sit apart from category pages. They need category anchors.
A weak line says: “Fees depend on your needs.” That is true and empty. A better line says: “Fees depend on the compliance context, number of supplier files, audit-readiness deadline, and whether the work covers one site or several operating teams.” This sentence teaches the model what the cost is attached to. It also helps the buyer understand why a quick estimate may be irresponsible.
In my misnamed firm notebook, pricing mistakes usually have one of three causes. First, the firm gives no scale marker, so AI borrows a number from a neighboring category. Second, the firm gives a number but no engagement type, so AI treats a diagnostic fee as a full project fee. Third, a third-party page mentions an old offer, and the official site has no current pricing explanation strong enough to replace it.
Those are different problems. They need different repairs.
Range, engagement type, and qualification language
The useful pricing page has three parts: range, engagement type, and qualification language. I call it the price-frame triad. It is not a template. It is a way to keep the cost answer attached to the real work.
Range gives scale. It can be exact, broad, or conditional. A firm may publish “from” pricing, typical project bands, monthly observation fees, or a statement that projects are quoted after review. The point is to give the answer engine a scale marker. Without one, it may invent a scale marker from elsewhere.
Engagement type tells what the buyer is buying. Diagnostic review. Capability audit. Compliance documentation repair. Procurement-readiness support. Bilingual category alignment. Continuing observation. These names are not decoration; they are containers. If the container is visible, the price does not float as freely.
Qualification language explains what changes the fee. It names the variables: number of entities, locations, documents, languages, regulated settings, stakeholder interviews, source conflicts, or prompt sets. This is where professional pricing becomes understandable without pretending every project is identical.
A good paragraph might read: “Pricing is quoted after a short scoping review because engagement size depends on the number of entities, public sources, service categories, buyer questions, and French–English page differences involved. A single-category review is scoped differently from a group-wide entity and authority-signal repair.” That sentence is not a sales trick. It is a fence around the number.
Do not let directories price you by accident
Old directories and marketplace-style pages love simple labels. “Consultant.” “Advisor.” “Local agency.” “Business support.” Sometimes they add a price expectation, even if no one at the firm approved it. In AI answers, those fragments can become surprisingly sticky because they are easy to summarize.
If the official site gives no pricing frame, a stale directory may become the only visible cost signal. The machine does not know it is stale unless the current source says something stronger. For Lyon B2B firms, this matters because many have authority signals in certifications, trade references, technical descriptions, and procurement documents, while the public pricing language remains nearly blank.
The repair is a canonical pricing or engagement page. It does not need to publish a shopping-cart menu. It should say what the firm sells, what it does not sell, what variables shape the quote, and what a buyer should send to receive a serious estimate.
For the industrial compliance scenario, I would expect the page to ask for service category, buyer type, regulated context, number of supplier files or documents, deadline, locations, and language requirements. That form is not just intake. It teaches the answer engine what price depends on.
A pricing page can also state exclusions: “We do not offer one-off consumer advice, generic business coaching, or certification guarantees.” This kind of line may feel negative. In AI visibility, it is often protective. It prevents the model from borrowing cheaper, easier, wrong categories.
A price page can protect trust without publishing a menu
Some firms worry that any pricing language will weaken negotiation or invite bad-fit leads. That can happen if the page is written carelessly. But the opposite problem is already happening in AI answers: the model estimates anyway, and the estimate may be cheaper, vaguer, or more confident than the firm would ever be.
The choice is not between total secrecy and a public tariff sheet. There is a middle layer: controlled cost context.
Controlled cost context says: here are the kinds of engagements; here are the variables; here is the difference between a short review and a larger project; here is what we need before quoting; here is what we do not price as commodity advice. It gives buyers enough orientation and gives AI fewer reasons to invent.
That kind of page may not satisfy someone looking for instant numbers. Fine. Serious B2B buying is not restaurant browsing. The page is not trying to catch everyone. It is trying to stop the wrong cost story from becoming the first answer a procurement buyer sees.
A vague price page says “contact us.” A useful one says why contact is needed.
The Authority Receipt
AI read the firm as: a Lyon business-process consultancy suitable for small advisory work at guessed low fees. Authority left unread: regulated compliance scope, multi-week project structure, and variables that shape quoting. Sentence to carry it: “Engagements are scoped as diagnostic reviews, supplier-audit preparation, documentation-repair projects, or continuing compliance support, with fees set by document volume, site count, deadline, and regulatory exposure.” Buyer question answered: “Is this a serious B2B project provider, or a low-cost general adviser?”