The 6-point AI-readiness checklist every UK retailer needs before Christmas.
By Christmas, more of your customers will ask an AI what to buy than you think, and the AI will answer with a shortlist. Universal Cart, Google's AI Mode, AI Overviews and the non-Google assistants like ChatGPT and Perplexity all pick that shortlist the same way: from the data they can read about your products, not from how nice your website looks. If your data is thin, you are not on the list, and you never find out why.
The good news is that getting AI-ready is not a mysterious new discipline. It is six concrete things, most of which you can check yourself this afternoon. Six things every retailer needs in place before Universal Cart lands. Save this one, work through it, and you will be ahead of most of your competitors, who have done nothing about any of it yet.
1. Your product feed
Everything starts here, because the feed is what the AI reads first. Every product needs clean, accurate titles, correct categories, full attribute coverage, and stock and price kept up to date, because agents check availability before they recommend anything.
The newer opportunity is Conversational Attributes: extra Merchant Centre feed fields, live in the UK, built specifically for AI shopping (product Q&A pairs, popularity rank, related products, links to specs and manuals, variant grouping). These are exactly what an AI reaches for when it answers a shopper's question, and most UK retailers have filled in none of them.
Quick check: open your feed and look for blanks, "N/A" and missing GTINs. Then ask whether you have touched Conversational Attributes at all. Deeper reading: Conversational Attributes in Merchant Centre and the Universal Cart product-data homework.
2. Your product images
Agents pull your images to reason about the product, so bad images mean not chosen. You need a sharp, well-lit primary image on a clean background, plus lifestyle alternatives that show scale and context. Google's visual search actively favours images that show the product in real-world use over sterile pack shots, so a catalogue of white-background-only images is quietly holding you back.
Quick check: pick your ten best-selling products. Do they each have a clear primary image and at least one in-context lifestyle shot? If they are all plain pack shots, that is your first image job.
3. Your reviews and ratings
When an AI compares options, it weights review count and score heavily, because reviews are the closest thing it has to trust. A great product with three reviews loses to an average one with three hundred. This is not something you fix in an afternoon, which is exactly why it is worth starting now, before the season.
Quick check: how many reviews do your hero products carry versus your nearest competitor's? If you are behind, get an active review-generation programme running on Google and on your own product pages, with Trustpilot where it fits.
4. Schema markup on your site
Schema is the translation layer that lets machines read your web pages reliably, and it matters most for the non-Google agents, ChatGPT, Claude and Perplexity, that lean on it to understand your products. You want Product schema (price, availability, brand, GTIN), Review schema (rating and count), and FAQ schema where it fits.
Quick check: run a key product page through a schema validator, or ask your developer what structured data is on it. If the answer is "none" or "not sure", that is a cheap, high-impact fix. Deeper reading: Product, Review and FAQ schema for AI recommendations.
5. llms.txt on your site
If schema helps AI read individual pages, llms.txt helps AI find and understand your site as a whole. It is a simple file, the AI-era cousin of robots.txt, that makes your products discoverable to the non-Google assistants. It pairs naturally with Conversational Attributes to give you coverage across both Google and the wider AI landscape.
Quick check: try loading yoursite.co.uk/llms.txt. If it 404s, you do not have one. Deeper reading: llms.txt explained.
6. Price competitiveness
AI agents compare prices ruthlessly, on data they trust, and being a few pounds cheaper often wins the recommendation outright. The risk is not just being expensive; it is being quietly undercut on a key product and never knowing, because the agent simply stops surfacing you. Human shoppers forgive a small price gap once they like a brand. An agent has no such loyalty.
Quick check: do you know, today, whether you are undercut on your top twenty products? If the honest answer is no, a daily price-monitoring layer is the cheapest insurance you can buy against silent AI demotion. See our Price Monitor.
How to use this checklist
Do not try to do all six at once. Work top-down: the feed and images decide whether you are even considered, so start there, then reviews and schema, then llms.txt and price monitoring. Each point compounds the others, and none of them is a big project on its own.
It is also worth understanding what is coming, so the work feels less abstract. Universal Cart is the buy-inside-Google shift these six points prepare you for, AI Overviews are already changing who gets the click, and our guide to the 2026 search shift explains how the AI decides who to show in the first place.
What we do, and where to start
If you would rather know exactly where you stand before you start fixing things, that is what our AI Visibility Report is for. It tests where you actually show up when a shopper asks an AI for a product like yours, whether you are being cited, and who is getting recommended instead, so you know which of these six points is costing you the most. Our Analyser then keeps the feed, price and performance side honest month to month.
Work through the six points and you will be genuinely ready for the AI-led Christmas. Ignore them and the AI will simply recommend the retailer who did.
Frequently asked questions
Do I need all six, or can I pick a couple?
Start with the feed and images, because they decide whether the AI considers you at all. But the six compound: reviews and schema and price all matter once you are in the running. Treat it as an order of work, not a menu.
Is this only about Google?
No. The feed, images and Google-side work help you inside Google's AI surfaces, while schema and llms.txt make you readable to the non-Google assistants like ChatGPT and Perplexity. The checklist deliberately covers both.
Why before Christmas specifically?
Because Q4 is when the buying happens, and some of these (reviews especially) take weeks to build. Doing the work in the summer means you are ready when it counts, rather than scrambling in November.
