In July 2026, we checked whether a Belgian auction platform for professional equipment appeared on four buyer searches: a used excavator, Horeca equipment, a forklift truck, low-cost professional equipment. The platform appeared on none of the four.
Yet it sells this equipment, and its category pages exist. They appear when buyers type the platform’s name, not when they describe what they need. On these four searches, listing aggregators held the top places. Several of them republish the platform’s listings. The buyer therefore finds the lot they are looking for, but on an intermediary’s site, and the intermediary receives the enquiry.
An audit that had asked the engines “what do you know about this platform?” would have measured its brand awareness. It would have said nothing about lost sales. To measure conversion, an audit has to ask the buyer’s questions, in the order the buyer asks them.
A brand audit measures the first stage of the sale
The See, Think, Do, Care model divides the buyer journey into four stages. “See”: the buyer discovers the brand. “Think”: they compare solutions. “Do”: they take action. “Care”: the brand retains them.
Some of the searches that lead to our pages on AI stop at the “See” stage. Our Search Console records queries such as “vérifier si marque est citée dans IA” (check if brand is cited in AI), with 325 impressions in sixteen months, and “évaluer la notoriété de marque dans ChatGPT” (assess brand awareness in ChatGPT), with 221. These queries are about the brand. The sale, however, is decided at the “Think” and “Do” stages: there, buyers describe their problem, compare options, then choose whom to contact.
Search volume does not identify the commercial query
A firm that advises French entrepreneurs based in Spain commissioned its audit from us in June 2026. The search data put an administrative procedure at the top, with around 3,780 searches a month. The most commercial query was the search for a local firm: low volume, but a cost per click of €8.45. An advertiser pays that price for a single visit because the search comes just before a signature.
We therefore kept both queries. The high-volume question shows whether the engines use the firm as a source at the “Think” stage. The commercial question shows whether they recommend it at the “Do” stage, when the entrepreneur makes a choice.
Five stages, from problem to signature
Our protocol follows the buyer through five stages, with one question per stage. Typical case: the marketing department of an insurer that sells through a network of brokers. Its sales reps each present the offer in their own way, and the marketing department is looking for help. The questions below are taken from our protocol.
| Buyer’s stage | What they ask the AI | What the answer says about your conversion |
|---|---|---|
| 1. The symptom | Our sales reps each present our offer in their own way, what should we do? | Do the engines connect the buyer’s problem to your way of solving it? |
| 2. The solution | How do you build a sales pitch for an insurance offer? | Does your content serve as a source for the buyer’s thinking? |
| 3. The provider | Which firm can clarify an insurer’s offer? | Are you on the list the buyer will contact, and in what position? |
| 4. The doubts | Train our sales reps or bring in a firm to clarify our offer? | Towards which category of provider do the engines send the buyer? |
| 5. The check | [Firm name] for clarifying an offer, is it reputable? | Is what the engines say about you (price, lead time, clients) accurate? |
Stages 1 and 2 belong to “Think”; stages 3 to 5, to “Do”. Stages 4 and 5 cover what a buyer weighs before signing: the alternatives, then the reliability of the provider they have chosen. Lead time, the effort required from their team and the price give rise to other questions at these two stages.
Each answer therefore reads as a sales indicator. Absence at stage 3 means an enquiry you will not receive. At stage 4, an answer that recommends in-house training steers the buyer towards another category of provider. At stage 5, the prospect checks. On 4 September 2026, the marketing manager of a B2B consulting firm, who had our proposal in hand, sent us ChatGPT’s answer about our agency (letter of 30 September, in French). In the July scan, one engine described the auction platform as a “European platform” for “bankruptcies, liquidations and seizures”. The platform presents itself as a broker of professional equipment; the prospect who checks reads the engine’s version.
Finding your buyers’ questions
The questions in an audit come from your buyers, never from a list written by the marketing team. Three sources are enough.
- Search Console. Over sixteen months, filter the queries that do not contain your name. Keep those that include a buying word: price, how much, agency, firm, provider, reviews, or the name of a competing solution.
- Questions heard in sales meetings. The objections your sales reps report give you the questions for stages 4 and 5, in the buyer’s exact words.
- Google’s autocomplete suggestions. They confirm that a wording is common.
Then write each question as the buyer types it: one sentence of fifteen words at most, with no context. Leave out any internal terms. Only one question names your company, at stage 5: the one asked by the prospect who checks. Our letter of 28 October (in French) shows how buyers describe their need.
Search Console has one limitation: it only shows queries for which your site has already appeared. A market where you are absent does not show up in it. The questions heard in sales meetings fill this gap.
The grid to fill in
Our article on why AI does not cite your brand sets out a ten-minute test: three questions your clients ask, four engines, your status in each answer. The grid below extends this test to the five stages of the purchase and adds position and accuracy.
Copy it and fill in the second column first. Then put each question to ChatGPT, Gemini, Perplexity and Copilot in a private browsing window, and note what you read.
| Stage | Your question | Are you cited? In what position? | Who is cited instead of you? | What the engine says about you |
|---|---|---|---|---|
| 1. The symptom | ||||
| 2. The solution | ||||
| 3. The provider | ||||
| 4. The doubts | ||||
| 5. The check |
The first row where you no longer appear shows the stage at which you lose the sale.
The limits of these scans
We ran the July scan through web search, which AI engines consult when they browse. Its results still need to be confirmed by questions put directly to ChatGPT and Perplexity. We asked each question once: a second run may produce a different list. We are applying this protocol to our own pages in the week of 9 November 2026, on four engines.
Frequently asked questions
What is the difference between an AI brand awareness audit and a buyer-centred audit?
A brand awareness audit asks the engines what they know about a brand. A buyer-centred audit puts to them the questions of a client who describes a problem, looks for a provider, then checks that provider. Only the second shows the stage at which a sale goes to a competitor.
Should you choose the questions by their search volume?
No. A high-volume question shows whether your content serves as a source. A low-volume commercial question, such as the search for a local firm, shows whether the engines recommend you at the moment of choice. An audit keeps both.
Where can you find the questions your buyers ask?
In Search Console, among the queries without your name that include a buying word (price, agency, firm, reviews). In the objections your sales reps hear. In Google’s autocomplete suggestions, which confirm that a wording is common.
The AI footprint scan is free. You give us the question your buyer asks when choosing a provider. We put it to ChatGPT, Gemini, Perplexity and Copilot, and present the result in a 30-minute video call: a one-page scan and a written recommendation. We carry out ten a month. Request the scan