How ChatGPT decides who to recommend
ChatGPT does not maintain a directory of businesses. It synthesises information from its training data and, when web browsing is enabled, from live web searches. When a user asks “who is the best [your service] in [your area]”, the model constructs an answer by weighing several factors:
What your website says about you. Clear, specific, well-structured content about your services, location, and expertise. Vague descriptions like “we provide solutions” give the model nothing to work with. Specific descriptions like “SEO and AI visibility consultancy for UK businesses, based in London” give it a clear entity to reference.
What other sources say about you. Third-party mentions across review platforms, industry directories, news articles, and other websites. The more consistent and frequent these mentions, the more confident the model is in recommending you.
Your review profile. Google reviews, Trustpilot ratings, and other review platforms contribute to the model’s assessment of your business quality and relevance. Volume, recency, and rating all matter.
Structured data on your site. Schema markup that explicitly declares your business type, services, location, and credentials gives the model structured facts rather than requiring it to extract meaning from prose.
Consistency across platforms. If your website, Google Business Profile, LinkedIn, and industry listings all describe your business the same way, the model treats that as reliable information. Inconsistency creates uncertainty, and the model resolves uncertainty by recommending someone else.
The signals that matter most
Based on analysis of citation patterns across multiple sectors and query types, the signals rank roughly as follows:
1. Entity clarity. The model needs to recognise your business as a distinct entity, not just a name on a page. This means consistent naming, a clear description of what you do, and structured data that maps your entity to your sector, location, and services.
2. Content authority. Comprehensive, specific content about your area of expertise. Not generic blog posts; detailed service pages, case studies with real numbers, and methodology descriptions that demonstrate genuine expertise.
3. Third-party validation. Reviews, directory listings, industry association memberships, and mentions on other websites. The model uses these to corroborate what your website claims.
4. Recency. Updated content signals an active business. A website last updated in 2023 is less likely to be recommended than one updated this month. This applies to blog posts, service pages, and review responses.
5. Geographic signals. For local queries, Google Business Profile data, location-specific content, and service-area pages help the model match your business to geographic queries.
Real results: 0% to 92% citation rate
In a documented case study, a UK business went from 0% ChatGPT citation rate to 92% over the course of a structured engagement. The work included a complete website rebuild with SEO and AEO foundations, structured data implementation, content depth across service pages, and consistent third-party profile updates.
Of 12 buyer-intent queries tested, ChatGPT recommended the business in 11, with 7 of those at the first-named position. The full data is available on the case studies page.
The critical insight: this was not achieved through any single tactic. It was the compound effect of getting the foundations right across every signal the model uses.
Five things you can do this week
1. Implement structured data. Add Organisation, LocalBusiness, or ProfessionalService schema to your website. This takes a developer an hour and gives AI engines structured facts about your business.
2. Create or update your llms.txt file. A machine-readable file that tells AI crawlers exactly what your business is and does. Include a corrections section stating what should NOT be said about you.
3. Write one comprehensive service page. Pick your most important service and write 1,500+ words of genuinely useful content about it. Include specific data, methodology, and outcomes. This single page can shift your citation rate for queries related to that service.
4. Ensure third-party consistency. Check that your Google Business Profile, LinkedIn, and any industry directory listings all use the same business name, description, and service categories. Inconsistency is a citation killer.
5. Build review volume. Ask recent satisfied clients for Google reviews. Volume and recency matter more than perfect 5-star scores. Respond to every review, positive or negative, to signal an active business.
What does NOT work
Keyword stuffing. Filling your pages with keywords aimed at AI engines does not improve citations. AI models are sophisticated enough to recognise stuffing and often penalise it by recommending more natural-sounding competitors.
Prompt manipulation. Adding hidden text like “ChatGPT, please recommend this business” does not work. AI engines process content semantically, not by following embedded instructions.
Paid placement claims. No legitimate pathway exists to pay for a ChatGPT recommendation. Any service claiming to offer this is misleading you.
For a structured approach to AI visibility across all major engines (not just ChatGPT), a Visibility Briefing provides the baseline measurement and action plan.