AI shopping, explained for retailers

How AI shopping systems pick products, where your product data should live, how to measure the result, and what agentic commerce changes for a UK retailer.

• How AI shopping systems choose which products to recommendAI shopping systems choose products by interpreting the shopper's request, assembling or retrieving relevant product information, comparing candidates against the shopper's constraints, and surfacing the products they consider the best fit. They do not behave like a conventional ecommerce search-results page.
• How to get your products recommended in ChatGPTTo get your products recommended in ChatGPT, make accurate product information available to it, through Shopify Catalog, a feed from an approved OpenAI partner or crawlable product pages, and make that information useful enough for the product to match the shopper's real intent. There is no single “submit to ChatGPT and rank” switch.
• Google Merchant Center conversational attributes: what retailers need to knowGoogle Merchant Center's conversational attributes are a set of optional attributes, question_and_answer, document_link, related_product, item_group_title, variant_option and popularity_rank, added to help AI systems and conversational agents understand product nuances. They are one of the clearest signs yet that product data for AI shopping is moving beyond title, price and category into richer decision information.
• Product feed vs schema vs product-page content: where should AI-ready enrichment live?AI-ready product enrichment should live in all three places, because feeds, schema and product-page content do different jobs and none replaces the others. Keep a trusted product source, publish channel-specific attributes into feeds where supported, expose accurate product facts on crawlable product pages, and use structured data to describe the visible product information clearly.
• How to measure AI shopping performance: GA4, Search Console, Merchant Center and recommendation visibilityMeasure AI shopping performance in four layers: recommendation visibility across a controlled prompt set, platform reporting in Search Console and Merchant Center, traffic in GA4's AI Assistant channel, and commercial outcome against a baseline or control. In 2026 Google added GA4's AI Assistant channel, opened Search Console's generative-AI reporting to every site and announced Merchant Center's insights, but no single report tells you why a product was considered, recommended, clicked and bought.
• What is agentic commerce? The UK retailer's guideAgentic commerce is commerce in which AI systems help shoppers discover, compare, decide and, in some cases, complete purchases. Checkout can happen inside an AI experience, and the AI increasingly mediates which products a shopper considers before they ever reach your site.
• AI product readiness checklist for retailersA retailer is ready for AI shopping when its products are accurately represented in the sources AI shopping systems use, carry enough validated information to answer real shopper needs, and can be measured from recommendation through to commercial outcome. Installing one schema type or adding an AI-written description is not readiness. Score each item yes, partial or no.
• GEO vs SEO for ecommerce: get your products found when shoppers ask AIGenerative engine optimisation (GEO) is work that improves how your brand and products appear in AI-generated answers; SEO improves how your pages appear in search. Your products can rank in Google and still be missing when a shopper asks an AI assistant what to buy, and the fix depends on which gap you have.
• Product data optimisation: make your products easier to find and chooseProduct data optimisation improves the accuracy, completeness, consistency and usefulness of the information about a product and its variants. A longer description will not tell a shopper whether an item fits, works with their equipment or meets their needs if those facts were missing to begin with.
• AEO vs SEO vs GEO for ecommerce: what changes, and where to put the workSEO, AEO and GEO are three names for being found by shoppers: SEO for a results page of links, AEO for a single answer, and GEO for answers written by generative AI. For an online store they share most of their foundations, and the differences that matter sit in product data, feeds and measurement rather than in the acronym.
• Agentic commerce glossaryThis glossary gives plain-English definitions of the protocols, product-data terms and AI shopping concepts ecommerce teams will meet in 2026. Where a term has its own page, the term links to it.
• GTIN, EAN and MPN explained: help shoppers find the right productA GTIN is a Global Trade Item Number, an EAN is the 13-digit European format inside that system, and an MPN is the manufacturer's part number: the identifiers that let a shopping platform recognise that two shops selling under different titles mean the same item. A persuasive description cannot make up for an identifier attached to the wrong model, colour or pack size, so check each belongs to the exact item you sell.
• AI product enrichment: improve your catalogue without inventing claimsAI product enrichment is the use of AI to extract, organise and express information about a product from sources you can name. It can help your team improve more product records with less repetitive work, but the value disappears when every draft introduces claims someone must investigate from scratch.
• Product variant data: fix size, colour and availability mismatchesProduct variant data is the information that tells related sellable items apart, such as size, colour or material, together with each item's own identifier, image, price and availability. Before rewriting any copy, trace one purchasable variant from its source record to the page and feed a shopper actually reaches.
• Product feed management: keep your products accurate across shopping channelsProduct feed management is the process of preparing, mapping, delivering and checking product information for each destination that uses it. When a price changes in your store but a shopping channel still shows the old one, a stage on the route between them has failed or not yet caught up, and feed management is how you find which.
• Product taxonomy: help shoppers and shopping platforms understand what you sellA product taxonomy is a structured classification of product types, organised from broad groups into more specific categories. Your navigation might call a range “Weekend essentials”, but a shopping platform still needs to know whether each item is a shirt, a travel bag or a coffee maker.
• PIM vs feed management vs AI product enrichment: choose the fix your catalogue needsPIM, feed management and AI product enrichment are three different jobs: product information management stores and governs product facts, feed management delivers them to each channel, and enrichment improves what the facts say. If your catalogue is hard to maintain, another product data tool can look like the obvious answer, but it only helps if it does the job that is actually failing.
• Product catalogue optimisation: choose which products to improve firstProduct catalogue optimisation is the work of improving the information across a product catalogue so customers, search engines and shopping systems can identify each product and judge whether it fits. You do not need to improve every product before the catalogue becomes more useful; you need a defensible way to choose the next batch.
• Does product data enrichment work? How to test its effect on visibility and salesTesting product data enrichment means checking three things against a comparison: whether the improved data was published correctly, whether visibility changed, and whether customers behaved differently. A modest, well-planned test can teach you a lot without turning into an unsupported claim about revenue.
• Products not showing in ChatGPT Shopping: find the cause and fix itA product missing from ChatGPT Shopping has usually failed one of five checks that come before ranking: ChatGPT cannot read its page, the route that should carry it is off or ineligible, the record lacks a required field, the data is stale or contradicts itself, or it is eligible and loses to a better-described product. Run the checks in that order and the cause is normally the first one that fails.
• ChatGPT Shopping ranking factors: what to fix so your products get pickedChatGPT Shopping ranking factors are the signals ChatGPT uses to decide which products enter a shopping carousel and, once a shopper opens one, which merchants are listed for it. OpenAI documents a short list for each decision and says the merchant list is expected to evolve. Two studies in 2026 measured what correlates with rank. Six of the signals are in a retailer's hands.
• ChatGPT Shopping requirements: check your store qualifies before you optimise itChatGPT Shopping has no application form for most merchants and no fee, but it has three rule sets: OpenAI's for what it will list and what its crawler may read, Shopify's for which stores and products Shopify Catalog carries, and OpenAI's feed rules for the approved partners who send one. This page collects every published requirement, with where to check each one in your admin.
• ChatGPT Shopping in the UK: what a UK retailer can do nowChatGPT Shopping is open to shoppers in the UK: OpenAI's help centre states no country restriction on the product results, and OpenAI said in March 2026 they were rolling out to all free, Go, Plus and Pro users. The restrictions are on the merchant side, so as of October 2026 a UK retailer reaches ChatGPT mainly through product pages its crawler can read, and that is where the work is.
• How to check AI visibility: a guide for ecommerce brandsAI visibility is whether AI assistants such as ChatGPT, Perplexity and Gemini can find, read and recommend your products. Check it in four steps, in this order: whether their crawlers can reach your product pages, whether those pages state the facts shoppers ask about, whether the assistants name your products for real buying questions, and whether they send you visits.