What’s actually happening when an assistant recommends one product over another.
When a customer asks ChatGPT “what’s the best espresso machine for a beginner?” something very different from a Google search is happening. There’s no list of ranked links. The assistant analyses, weighs options, and produces a shortlist. One brand makes it. Most don’t.
For ecommerce teams, the question is obvious: what decides who gets recommended?
The competitive surface just got smaller
Traditional search returns ten blue links. AI assistants typically surface two to five products per query. Missing that shortlist doesn’t mean ranking lower — it means not being considered at all. The stakes of product visibility have changed fundamentally.
What AI systems need to make a recommendation
AI assistants are not matching keywords. They are reasoning about fit. When a customer asks about running shoes for flat feet, the assistant is trying to answer: does this product actually address this need? Can I say that with confidence?
Products that are easy to reason about get recommended. Products that are vague, inconsistent, or poorly structured get skipped — not because they’re worse, but because the assistant can’t evaluate them confidently.
The clarity gap
Most product pages were written for human browsers, not machine reasoning. A description like “premium quality, perfect for everyday use” gives an AI assistant almost nothing to work with. It can’t determine use case fit, compare it against alternatives, or include it in a specific recommendation with confidence.
The brands winning in AI discovery are the ones whose products are specific, structured, and consistent — making it easy for AI systems to say “yes, this fits” without uncertainty.
Why clusters matter
AI assistants rarely recommend one product in isolation. They build shortlists — two to five options that fit the query from different angles. To be included, your products need to be part of a well-defined discovery space, not floating in isolation.
That’s why the unit of AI visibility isn’t the product page. It’s the discovery layer that organises your products around the questions customers actually ask.
What this means for your store
The brands showing up in AI recommendations today didn’t get there by accident. They have clear product data, well-organised discovery coverage, and content that makes it easy for an assistant to say “this fits.” That infrastructure can be built. But it takes more than SEO.
Geoffy helps ecommerce teams build the discovery layer that gets products into AI recommendations. See how it works →
Frequently asked questions
How do AI assistants choose which products to recommend?
They reason about fit — evaluating how well a product matches the user’s specific question based on the clarity and completeness of available product information.
Why do some products get recommended and others don’t?
Products that are easy for AI systems to interpret and evaluate confidently are more likely to appear in recommendations. Vague or inconsistently structured products give assistants less to work with.
Is this the same as SEO?
No. SEO is about ranking pages. AI discovery is about whether assistants can understand and recommend your products within generated answers.
What is GEO’s role here?
Generative Engine Optimisation (GEO) is the practice of structuring product content so AI assistants can interpret and recommend it confidently. Learn more about GEO →
About the author
Anthony Gale is Co-Founder of Geoffy, a Generative Engine Optimisation platform for ecommerce brands. He has spent more than two decades working in ecommerce and digital growth, helping retailers adapt to major shifts in online discovery.