TL;DR: Nobody outside the AI companies can tell you exactly how an assistant picks a store, and anyone who promises a recommendation is guessing. What is known is the groundwork: be indexed, keep product data accurate in your pages and your Merchant Center feed, collect real reviews, and write your policies and answers as plain text. Google says its AI features need nothing beyond that.
The owner of a small online yarn store asks an AI assistant, as a test, "where can I buy undyed merino yarn for sock knitting online?" The answer names two large retailers and a marketplace. Her store, which has sold that yarn for six years, is not there. Her first thought is that AI search has buried small shops for good.
It is a fair worry, and the honest answer to how a small online store can get recommended by AI assistants is partly "you cannot control it" and partly "you control more than you think". This article separates the two. It is part of our guide to marketing strategies for a small Shopify store.
How this was made: written by Miriad with AI assistance and checked against the sources listed at the end.
What makes an AI assistant recommend a small online store?
No one outside the AI companies knows the exact selection rules, but the inputs are visible: what the assistant can read about your store and products. Google says its AI features have no additional technical requirements. Pages must be indexed and snippet-eligible, and Merchant Center and Business Profile information should be current. The rest is content quality.
That comes from Google's own page on AI features and your website, which says AI Overviews and AI Mode have "no additional technical requirements". In other words, the work that makes a store readable for search is the same work that makes it readable for Google's AI features. Other assistants publish less about how they choose, so treat the same groundwork as your best available bet rather than a formula.
What an assistant can read about a small store falls into four groups: the store's own pages, its product data in feeds and structured data, what others say about it (reviews, mentions), and how clearly it states facts like shipping and returns. The checklist below follows those four groups.
Is AI search big enough to worry about yet?
It is growing quickly from a small base. Adobe measured traffic to US retail sites from generative AI tools rising 693.4% year over year in the 2025 holiday season, and added that the user base "remains modest" (Adobe). Worth preparing for, with work that also helps regular search, but not worth rebuilding the store around.
The figure measures referral traffic to US retail sites in Adobe's data, not small stores specifically. Read it as direction. The practical conclusion is that nothing in the checklist below is AI-only work: each item also helps Google Shopping, organic search or the buyer reading your page.
The GEO checklist for a small online store
GEO (generative engine optimisation) for a small store comes down to four groups of work: be crawlable and indexed, keep product data complete and consistent, show real proof such as reviews, and write the answers buyers need as plain text on the site. Each item below is something a one- or two-person team can check.
1. Crawlable and indexed
2. Product data complete and consistent
3. Real proof
4. Answers as plain text
The plain-text policies matter to people too. In 2025, 8% of EU e-shoppers said information on warranties and legal rights was hard to find, per Eurostat. Writing it clearly helps buyers first and any assistant reading the page second.
A worked example of the checklist
Score one point for each ticked box, then fix the unticked items in the group with the fewest points first. Groups 2 and 4 usually have the quickest fixes for a small store, because they are about data and text you already have. Re-score after a month and keep a dated copy each time.
Example (illustrative, not a real case): the yarn store from the introduction scores itself on the 13 boxes above.
| Group | Boxes | Before | After one month |
|---|---|---|---|
| Crawlable and indexed | 3 | 3 | 3 |
| Product data | 3 | 1 | 3 |
| Real proof | 3 | 1 | 2 |
| Answers as plain text | 4 | 1 | 4 |
| Total | 13 | 6 | 12 |
The owner added merchant listing fields, connected the feed, wrote a shipping and returns page in plain text, added fibre, weight and care answers to her 15 bestsellers, and started asking every buyer for a review. Reaching 50 reviews will take months. Whether an assistant now names the store is still outside her control; what changed is that there is now something specific and accurate for it to find.
What can't a small store control?
You cannot control which sources an assistant was trained on, how it weighs large retailers against small shops, or whether it names any store at all for a given question. You also cannot buy a recommendation through a special file or markup. Spend the time on the checklist and on being mentioned by real customers.
Mentions outside your site are the part of the picture you influence slowly: a customer's review on Google, a knitting blogger's guide that lists your yarn, a forum answer that names your store. They come from doing the basics well over time. Paying for fake mentions or reviews breaks platform rules and, in the US, the FTC's fake-review rule; the reviews article covers the safe route.
How do you measure AI-search visibility?
Measure in two ways: Search Console data for Google, where AI-feature traffic is counted under the "Web" search type, and a dated manual test of the same buyer questions in two or three assistants each month. Neither is precise. Together they show whether the store is showing up more often over time.
According to Google Search Central, traffic from AI features appears in Search Console's Performance report under the "Web" search type, together with regular results, so you cannot split it out exactly. Watch impressions and clicks on the pages you improved.
For the manual test, keep a sheet: date, assistant, question, stores named, whether yours appeared. Ask the same five questions each month. It is anecdotal, but it is your own record rather than someone else's claim.
For the technical side, run the Shopify store SEO checklist. For the full plan, return to the Shopify store marketing guide.
If you want a guided method for the GEO checklist, from product data to pages that AI assistants can quote, we are building the SEO & GEO Operator Pack. Get notified when it launches.
FAQ
Can I pay to be recommended by AI assistants?
Not in the sense of buying a place in an assistant's answer through a file, markup or service that promises it. Some assistants show ads or shopping results, which follow their own rules, but an organic recommendation cannot be bought. Be wary of anyone who promises AI visibility for a fee. The reliable work is accurate product data, real reviews and clear pages.
Do I need an llms.txt file or special markup for AI search?
Google says its AI features have no additional technical requirements, so for Google the answer is no. Standard product structured data, an indexed site and a current Merchant Center feed are what it asks for. Other assistants publish less, and there is no public evidence that a special file makes them recommend a store. Spend the time on content and product data first.
Why do AI assistants name big retailers instead of my store?
Large retailers have more pages, more reviews and more mentions across the web, so there is more for an assistant to find about them. You cannot match their volume, but you can be more specific: detailed answers about your products, clear policies and real reviews. For narrow questions in your niche, a specific and accurate page has a better chance than a general one.
Sources
- AI features and your website — Google Search Central, 2025.
- Holiday shopping season drove a record $257.8 billion online — Adobe, 2026.
- Product (Product, Review, Offer) structured data — Google Search Central, 2025.
- About free listings — Google Merchant Center Help, accessed 2026.
- About Product Ratings — Google Merchant Center Help, accessed 2026.
- E-commerce statistics for individuals — Eurostat, 2026.