Every day, thousands of executives and marketing managers type into ChatGPT questions like "which SEO agency to choose in Belgium" or "best project management software for SMEs". ChatGPT responds with 3 to 5 recommendations — and your business is either included, or it isn't.

According to an analysis covering 2,847 audits conducted in 2026, approximately 78% of French and Belgian companies appear in no responses generated by the main LLMs. They're invisible to a growing share of their prospects — without even knowing it.

This guide explains how to check your ChatGPT visibility in 15 minutes, interpret the results, and activate the right levers to appear.

Why ChatGPT has become a B2B acquisition channel

ChatGPT now surpasses Bing in daily traffic volume. One in five internet users worldwide uses it every month. And according to Forrester 2026 data, 90% of B2B buyers now integrate generative AI tools into their purchasing journey — half of whom start directly in ChatGPT rather than on Google.

This isn't a marginal phenomenon. It's an acquisition channel structuring itself, with its own rules — and its own winners.

The fundamental difference from Google: ChatGPT doesn't return 10 blue links. It gives a synthesised response with 2 to 7 cited sources. If you're not in those sources, you don't exist for that prospect — even if you're on the first page of Google.

How ChatGPT works when answering a question about your sector

Before testing your visibility, you need to understand how ChatGPT chooses its sources. It relies on three distinct mechanisms:

1. Its training base

The base model was trained on data up to a cutoff date. Companies well-documented on the web before that date — via their site, press articles, LinkedIn mentions, sector forums — are in its memory.

2. ChatGPT Search (real-time web)

When the user activates web search or asks a recent question, ChatGPT queries the web in real time. According to an analysis of 730,000 conversations conducted in late 2025, the first question in a conversation is 2.5 times more likely to trigger a web citation than subsequent questions. Winning the first query is crucial.

3. Authority signals

ChatGPT cross-references your brand data with third-party sources: Wikipedia, LinkedIn, professional press, Crunchbase. LinkedIn is now cited in 14.3% of ChatGPT Search responses — it has become one of the strongest and most dynamic authority signals.

The 15-minute test: check your visibility now

Open ChatGPT (free or paid version) and test these 5 types of prompts, replacing the brackets with your actual situation.

Sector discovery prompts

These queries simulate a prospect who doesn't yet know you: "Which [agency / firm / provider] specialising in [your field] do you recommend in [your country / region]?" "What are the best companies for [your main service] for a B2B SME?" "I'm looking for an expert in [your field] for [typical client use case]. Any recommendations?"

Comparison prompts

These queries simulate a prospect in the evaluation phase: "Which players are recognised in [your sector] in [country]?" "Compare [your type of service] approaches: what distinguishes the best providers?"

Direct verification prompt

Test whether ChatGPT knows you: "What do you know about [your company name]?"

Interpreting the results

You appear with correct information ✅

Your company is in the training base or cited by third-party sources. Good starting point — but verify that the information is accurate and up to date.

You appear with incorrect information ⚠️

This is the riskiest scenario. ChatGPT may present your company with wrong positioning, incorrect services or an inaccurate geographic area. Priority: clarify your web content and third-party profiles.

You don't appear in sector recommendations ❌

This is the case for 78% of companies. This doesn't mean you're bad — it means you're not yet structured to be cited by an AI.

ChatGPT says it doesn't know you ❌

Your company is absent from its training base and the sources it considers reliable. This is the lowest starting point — but also the one with the greatest improvement potential.

The 6 reasons ChatGPT doesn't cite you

1. Your content doesn't directly answer questions

ChatGPT looks for pages that explicitly answer queries. A generic homepage isn't enough. You need content that answers word for word the questions your prospects ask.

2. You're not mentioned on third-party sources

LLMs trust sources that are themselves cited elsewhere. An article on your blog isn't enough if it's not picked up, shared or mentioned on other sites.

3. Your site isn't structured for AI

Without structured data (JSON-LD of type Organization, Product, FAQPage), ChatGPT doesn't precisely know what you do, for whom, and at what price. It will then prefer to cite a clearer source.

4. Your LinkedIn company profile is incomplete or inactive

LinkedIn is now one of the most cited sources by ChatGPT Search — and the most dynamically progressing. An active, well-completed LinkedIn company profile has become essential.

5. Your content is too recent or too scarce

LLMs favour sources with a consistent content history. A single article published last week has little chance of being cited. Regularity and volume matter.

6. You're not accessible to AI crawlers

ChatGPT uses GPTBot to index the web. If your robots.txt blocks GPTBot, or if your site doesn't have an llms.txt, you're inadvertently signalling that you don't want to be indexed.

The 5 concrete levers to appear in ChatGPT

Lever 1 — Create content that answers precise questions

Not generic articles about your sector. Articles that answer exactly the questions your prospects type into ChatGPT. Ideal format: title as a question + direct answer in the first 150 words + structured development.

Example for an HR agency: instead of "Our approach to recruitment", write "How to choose a recruitment firm specialising in finance in Brussels".

Lever 2 — Structure your data with JSON-LD

Add Schema.org markup on each page: Organization on your homepage, Product or Service on your offer pages, FAQPage on your FAQ page. These are direct signals that LLMs read to understand who you are.

Lever 3 — Activate your LinkedIn presence

Complete your LinkedIn company page to 100%, publish regularly (minimum 2 times per week), and ensure your service description is clear and precise. LinkedIn is indexed as a priority by ChatGPT Search.

Lever 4 — Get mentions on third-party sources

Submit your company to sector directories, participate in podcasts or interviews in your field, and target publications in professional media. Each external mention reinforces your authority in LLMs' eyes.

Lever 5 — Authorise AI crawlers

Check your robots.txt: GPTBot (ChatGPT), ClaudeBot (Claude), PerplexityBot and Googlebot must be authorised. Create an llms.txt file at the root of your site to describe your company, services and positioning in an AI-readable format.

How long before seeing results?

Visibility in ChatGPT isn't achieved overnight. Here are realistic timelines:

  • Structured data (JSON-LD, llms.txt): taken into account by ChatGPT Search within a few weeks
  • New targeted content: 4 to 8 weeks to be indexed and start being cited
  • Third-party mentions: progressive effect over 3 to 6 months
  • Training base: only during the next model updates (several months to years)

The opportunity window is progressively closing: dominant positions in AI responses are consolidating in 2026 and will be difficult to displace afterwards. Companies acting now are gaining a structural lead over their competitors.

Where to start concretely?

Start with the test described above. Note your results on the 5 prompts and identify which of the 4 scenarios you're in.

Then use our free scoring tool to get a structured assessment of your visibility on the 5 main AI engines — with a score per engine and the first personalised recommendations.

If you want to go further with a complete analysis including benchmarking of your direct competitors, our AI Diagnostic delivers a complete report within 5 business days.

Statistics cited in this article come from studies published between January and May 2026 by Forrester, Profound, ELMARQ and Winston Digital.