Find out whether ChatGPT, Perplexity and Gemini cite your brand, and course-correct: a GEO audit for a B2B SMB

Implementation example. This scenario describes a request we receive in this form. It is not a project delivered to a named client, and no results are claimed.

The context

An SMB sells to businesses: software, a service, equipment or a specialized offering. Its sales rely on inbound inquiries, trade shows and a partner network. The marketing team has one to three people, who handle the website, LinkedIn, content, campaigns and events.

The website was redesigned a few years ago. It presents the offerings, a few client references and a blog that is updated irregularly. Google rankings are monitored loosely, with Search Console.

The day-to-day problem

Sales reps hear more and more often: “I asked ChatGPT which providers were out there.” Sometimes the brand is cited, sometimes not, sometimes with false or outdated information: a discontinued offering, a price from three years ago, a competitor presented as the leader.

No one knows how to measure this. Traditional SEO tools do not show what AI engines answer. The team does not know which pages or external sources these engines use, or where to start.

What the SMB wants

“I want to know what AIs say about us when a prospect is looking for a solution like ours, why they cite our competitors, and what to do first to change that.”

The solution, step by step

1. Define the questions that matter. With the sales team, we list the questions a prospect actually asks before choosing a provider: “what are the best tools for…,” “how to choose a provider for…,” “how much does … cost,” “X or Y?”. We break them down by target audience and by language if needed.

2. Query the AI engines. The questions are asked to ChatGPT, Perplexity, Gemini and Google's AI answers, repeatedly, to account for variability in the answers. For each answer, we note whether the brand is cited, in what position, with what information, and which sources are mentioned.

3. Map the sources. We identify the pages and sites that feed the answers: pages on the brand's website, competitor sites, comparison sites, directories, press articles, forums, reviews. We spot false or outdated information and where it comes from.

4. Analyze the website. We check that the site is readable by AI engine crawlers (access allowed, server-side rendered content, no text locked inside images), that offerings, indicative prices and references are stated clearly, and that structured data describes the company and its offerings.

5. Prioritize the actions. The action plan ranks the workstreams by expected impact and effort: pages to create or rewrite (comparisons, use-case pages, FAQ), information to correct on the site and elsewhere, directory listings to complete, thought-leadership opportunities to plan.

6. Set up tracking. The same questions are asked again every month. A dashboard shows how citations, sources and the information picked up change over time.

What the AI does, what the team does

The AI asks the questions in bulk, compares the answers, extracts the sources and prepares the summaries. The team chooses the questions, validates the priorities and produces content or has it produced. The recommendations are reviewed by our consultants before being delivered.

The data and access required

access to Search Console and the site's analytics tool;
the list of offerings, target audiences and main competitors;
frequently asked questions reported by sales reps;
client references that can be published;
read access to the CMS if possible.

The project timeline

Days 1 to 3: scoping. Questions, target audiences, competitors, access.
Days 4 to 7: measurement. Querying the engines, mapping sources, analyzing the site.
Days 8 to 10: debrief. Prioritized action plan, presented to the team.
Then, if desired: production of priority content and monthly tracking.

The deliverables

the list of questions tested and answers obtained;
a brand presence score by engine and by question;
the map of cited sources, with the errors found;
the technical analysis of the site for AI crawlers;
a prioritized action plan;
the monthly tracking dashboard template.

The metrics to track

the share of questions for which the brand is cited, by engine;
the brand's position in the answer;
the number of false or outdated pieces of information picked up;
the share of cited sources that come from the brand's website;
inbound inquiries that mention a search via an AI.

These metrics are recorded during the audit, then tracked monthly if tracking is set up. No results are promised: no one controls what AI engines cite.

Limitations and prerequisites

AI engine answers vary from one time to the next: the measurement looks at trends, not at a single answer.
An audit changes nothing on its own: what matters is the content and corrections that follow.
The effects are measured over several months.

Indicative budget

The GEO audit starts from €1,000 excl. VAT, for 8 to 10 days. Content production and monthly tracking are quoted on request.

Frequently asked questions

How is this different from traditional SEO? SEO aims for a position in Google results. GEO aims to be cited in the answers written by AI engines. The two overlap, but the sources and formats that matter are not all the same.

Can you pay to be cited? No. Organic answers from AI engines cannot be bought.

Should AI crawlers be blocked or allowed? It is a choice to make with full knowledge of the implications. The audit shows what is currently blocked and what that means.

Can we get an initial measurement before the audit? Yes, GEO Radar, free and with no sign-up, gives an initial measurement on ChatGPT and Gemini in 3 to 5 minutes.

What next?

If you don't know what AI engines say about your company, start with GEO Radar, then the audit for a full measurement and an action plan.

Client challenge