Increasingly, shoppers aren't just typing a query into Google and clicking a blue link. They're asking ChatGPT to recommend a product, asking Perplexity to compare two options, or reading a Google AI Overview generated above the traditional search results. Generative Engine Optimisation (GEO) is the practice of making a site's content more likely to be found, understood, and cited by these AI tools when they generate an answer. This guide covers how these tools actually source their answers and what an ecommerce site can realistically do about it.
How AI Answer Tools Actually Source Information
AI answer tools generate responses from a mix of the underlying model's training data, real-time web retrieval, and, in some cases, structured data feeds. Google's AI Overviews, for example, draw on Google's search index and its Knowledge Graph, which includes structured product data submitted through Merchant Center and pulled from schema.org markup on retailer sites. Tools like Perplexity and ChatGPT with browsing enabled retrieve and read live web pages in a similar way to a search crawler, then synthesise an answer from what they find across multiple sources rather than presenting a single retailer's page.
This matters because it means visibility in AI answers isn't a separate, standalone discipline from SEO. It largely depends on the same underlying foundation: a page has to be crawlable, indexed, and clearly written for a retailer to have any chance of being cited, whether the eventual output is a ranked search result or a generated paragraph of text.
What Genuinely Seems to Help
Clear, Factual, Specific Content
AI tools tend to cite content that states facts unambiguously: specific measurements, materials, compatibility details, and comparisons, rather than vague marketing language. A sentence like "this jacket has a 20,000mm hydrostatic head waterproof rating and a 15,000g/m²/24hr breathability rating" is far more useful to an answer-generating tool than "engineered for all-weather performance," because the first is a fact it can extract and cite with confidence and the second is not.
Comparison and Specification Content
Content that directly compares products, whether your own range against itself or against competitors, tends to be well suited to how these tools construct answers to "which is better" or "what's the difference between" queries. A well-built comparison table with clear, accurate attributes is a genuinely useful format here, not just for readability but because it's structured in a way that's easy for a model to parse.
Structured Data
Accurate Product and Offer schema, kept up to date, supports how shopping-focused AI features (particularly Google's, given its access to Merchant Center data) represent a product's price and availability. This is the same structured data that supports traditional rich results, covered in Schema Markup for Ecommerce, so getting it right serves both purposes at once.
Third-Party Presence
Because tools like Perplexity and ChatGPT often triangulate across multiple sources rather than relying solely on a retailer's own site, a product's reputation elsewhere on the web (independent reviews, comparison articles on other sites, forum discussions) increasingly affects whether and how it gets mentioned. This is a natural byproduct of good content and genuine customer satisfaction rather than something to pursue as a separate campaign.
What GEO Doesn't Change
GEO doesn't replace the fundamentals covered elsewhere in this series: a technically sound site, unique product content, and accurate structured data remain the foundation regardless of whether the eventual destination is a search results page or an AI-generated answer. There isn't yet a reliable, standardised tool equivalent to Search Console for measuring AI answer visibility the way you can measure traditional rankings, which means monitoring this is currently more manual (testing queries directly in these tools and noting whether and how your brand appears) than it is for traditional SEO.
Being Realistic About This
This is a genuinely new and still-developing area, and we won't claim a specific, measurable result as if the science were settled, because with GEO, for anyone, it isn't yet. What we can say is that the underlying work, clear factual content, accurate structured data, and a technically sound site, is the same work that supports traditional SEO, so investing in it isn't a separate bet from investing in search visibility more broadly.
Update, 9 October 2026: Google has since published an official guide to optimising for AI features, and Search Console now has a Generative AI performance report that gives the first official measure of AI visibility. For checkout, see Google's Universal Commerce Protocol and ChatGPT's move to merchant apps.