The question comes up everywhere now: how do I get ChatGPT to recommend my business? The answer is less technical than most people expect, and slower than most people want.
There's no prompt hack. There's no tag to add. There's no shortcut that works cleanly. What there is: a set of things that create the conditions for AI models to mention you — and they're mostly things you'd want to do anyway.
Understand what the models are actually doing
AI models that respond to queries about products, services, and recommendations are drawing on two things: patterns in their training data, and (for models with retrieval) live search results. For a brand to be mentioned, it needs to exist in both — as a known entity with an established description, and as something that surfaces in the queries being retrieved.
That means presence. Not presence on your own site. Presence across the web.
The things that actually move the needle
Third-party coverage. Mentions in publications that have their own established credibility. Editorial comparisons. Category roundups written by people with no stake in the outcome. If the only source making claims about your brand is your own homepage, a model has nothing to cross-reference against.
Consistent category association. When articles about your category mention your brand alongside relevant use cases, models learn that association. If they don't, they don't. This is why niche editorial coverage — even from small publications — can matter more than a single mention in a major one.
Specific, factual content. Models prefer to cite brands they have clear, accurate information about. Vague mission statements don't give a model much to work with. A direct description of what you do, who it's for, and what makes it different does.
Structured, accessible information on your site. Pricing, features, use cases — presented clearly, not behind lead magnets or interactive sliders that models can't parse. If a competitor has transparent data and you don't, the model will often default to the competitor.
A real presence in user discussions. Forum threads, review platforms, community recommendations — when real users discuss problems and mention real solutions, models pick up on those patterns. This isn't something you manufacture; it's something that accumulates from having a product people actually find useful and talk about.
What doesn't work
Optimising for a specific model's preferences tends to backfire — the models update, and the tactics become obsolete. Thin coverage in low-authority directories adds noise without signal. Publishing AI-generated content at scale creates a lot of content that doesn't get cited.
The starting point
Run a structured audit first. Ask the major models your most important category queries — without mentioning your brand — and see who comes up. Then ask about your brand directly and note how it's described. The gap between those two results tells you what to work on.
The V-Score does this systematically and tracks it over time. Run your free AI visibility audit and get a baseline across ChatGPT, Claude, Perplexity, and Gemini.