A few years ago, "product discoverability" was almost entirely about Google Shopping and search ranking. Get your feed right, optimise your titles, earn some reviews — and you'd show up when buyers searched.
In 2026, product discovery has more layers. AI shopping assistants, generative search experiences, and AI-powered recommendation engines have created new pathways between customers and products. Many of these pathways depend on how well your product catalogue is structured for machine understanding — not just keyword matching.
Think about how a good librarian learns about books. They read the title, the back cover summary, the table of contents, the category label, and a few reviews. From all of that information, they can recommend the right book to the right reader — even for a request they haven't heard before.
An AI product discovery system works similarly. It reads your product data from multiple sources and builds an understanding of what your product is, who it's for, and how it compares to alternatives. The quality of its recommendations depends entirely on the quality of the data it can read.
The main sources AI agents draw from:
Structured data (also called schema markup) is code added to your product pages that explicitly tells search engines and AI systems: "This page is about a product. It's called X. It costs $Y. It has these reviews. It's available in these sizes."
Without structured data, AI systems have to guess this information by reading your page the way a human would — slower, less accurate, and more prone to misinterpretation. With it, the information is unambiguous.
The key schema types for ecommerce products:
Google's official structured data documentation provides the complete specification. Shopify, WooCommerce, and most major ecommerce platforms generate product schema automatically — but quality varies significantly. Check your own product pages using Google's Rich Results Test to see what's actually being detected.
Most product descriptions are written for humans — which is fine. But descriptions that work well for AI discovery share a common characteristic: they're semantically rich and specific.
What does this mean in practice? Instead of:
"Our best-selling beach mat is perfect for your summer adventures."
Write something like:
"The CGEAR Sand-Free Beach Mat (200cm × 120cm) uses dual-layer mesh technology to allow sand to fall through while keeping the surface clean. Suitable for beach, camping, and outdoor events. UV-resistant and machine washable. Weighs 800g with a compact carry bag included."
The second version answers the kinds of specific questions AI systems are asked: dimensions, how it works, what use cases it suits, weight and portability. An AI shopping assistant asked "find me a sand-free beach mat I can take camping" can confidently recommend the second product. The first? Not so much.
The test: Read your product description and ask: if a customer asked "is this product good for [specific use case]?" — could an AI confidently answer yes or no from your description alone? If not, it needs more specificity.
In both traditional Shopping and AI-powered discovery, the product title remains the most important single attribute. Think of it as the headline of your listing — it should contain:
Avoid filler words like "best," "amazing," or "great quality." AI systems optimised for relevance aren't looking for superlatives — they're looking for facts that help match the product to the right query.
AI shopping agents increasingly use reviews as a trust signal. A product with 200 verified reviews and a 4.7-star average will consistently surface above a product with no reviews — regardless of how good the title and description are.
The most effective approach for most ecommerce businesses:
For product visibility on Google's AI-powered surfaces — including AI Overviews and Google Shopping — your Google Merchant Center feed is the most direct connection.
A clean, complete, and regularly updated feed gives Google the structured product data it needs to surface your products in relevant AI-generated responses. Feed quality issues — missing attributes, outdated pricing, low-quality images — reduce visibility across all Google surfaces, not just traditional Shopping.
Google Structured Data for Products — Official schema documentation · Search Engine Land — AI search and product discovery · SEMrush Blog — Ecommerce SEO and AI visibility
Common questions about this topic.
An AI-ready product catalogue is one that AI agents — including Google's AI Overviews, Perplexity Shopping, and other AI tools — can read, understand, and confidently surface to users. This means each product has complete, accurate, and consistently structured data: descriptive titles, full specifications, correct pricing, Schema.org markup, and Merchant Center feed compliance. AI systems favour structured, trustworthy data over general website text.
AI-optimised product titles follow a specific structure: Brand + Product Type + Key Differentiator + Model/Size/Variant. For example, 'Ninja Air Fryer 5.7L Dual Zone AF400UK' is more effective than 'Ninja Air Fryer'. Include the terms customers use when searching, not internal codes. Titles should be 70–150 characters for Merchant Center and as descriptive as possible for on-site SEO.
Yes. AI agents assess data completeness when deciding which products to recommend. Products with populated attributes — colour, material, dimensions, weight, compatibility, care instructions — are ranked higher than products with sparse data. For Merchant Center, filling optional attributes like age_group, gender, material, and pattern significantly improves feed quality scores and AI discoverability.
For Merchant Center feeds, update at minimum daily to keep pricing and availability accurate. Google penalises feeds with mismatched prices (between feed and website) or products listed as in-stock when they're not. For product descriptions and attributes, update whenever product specifications change. Fresh, accurate data signals reliability to AI systems.
Yes — even active Shopping advertisers often have significant catalogue gaps that suppress performance. Common issues found in audits include disapproved products, missing GTINs, outdated titles, and incomplete attribute sets. Fixing these issues typically improves both Shopping ad impressions and organic AI visibility, since both use the same underlying feed data.
I audit product catalogues against AI visibility criteria — schema markup, feed quality, description depth, and review integration — and deliver a clear action plan.
See AI-Ready Catalogue Audit →