Ecommerce
How to get your products recommended by AI shopping assistants

More of your customers are asking an AI assistant what to buy before they ever search for it. They type "best waterproof hiking boots under $150" into ChatGPT, get three or four suggestions and click one. If your product isn't one of them, you never knew you were in the running. This post covers how those assistants pick products, what they can actually see about yours and the handful of fixes that make the biggest difference before the holiday rush.
Why this matters for your store right now
Last holiday season gave a clear signal. Adobe tracked a 693% jump in traffic from generative AI tools to US retail sites in November and December 2025 compared with the year before. Those visitors also converted 31% better than shoppers arriving from other sources.
Adobe didn't say how big that traffic is next to search and social, and it's likely still small. But it's the kind of traffic you want: people who've already asked their questions, compared a shortlist and shown up ready to buy.
Then in March, Shopify switched on what it calls Agentic Storefronts. If you sell on Shopify and your store is eligible, your products became discoverable inside ChatGPT by default, with Microsoft Copilot, Google AI Mode and the Gemini app on the same list. You didn't need an app or a new integration. So for most stores, the real question now is whether AI picks your products once it finds them.
How AI assistants decide what to recommend
None of these tools publish a ranking formula. What they do publish points in one direction. OpenAI's own guidance for merchants says products show up when shoppers are actively comparing options, that results are tailored to each shopper and that "richer data helps your products surface with more accurate information when shoppers compare options." Shopify's announcement says AI surfaces "favor structured data, with clean attributes and real-time accuracy."
Put plainly, an assistant recommends what it can describe with confidence. When someone asks for a gift for a runner who hates bulky gear, the assistant needs to know your product's weight, what it's made of, who it's for, what it costs today and whether it's in stock. If those facts are missing, buried in an image or written as vague brand copy, the assistant moves on to a product that spells them out.
That's the big shift from browsing. A person on your product page will put up with thin copy if the photos are great. An assistant reads the text and the data behind the page, so anything that's missing there simply doesn't count.
What the assistant sees when it looks at your products
When an AI tool pulls your product into an answer, it's working from a handful of sources. Each one is something you control.
Your product data
This is the title, description, price, variants, inventory and attributes in your store admin, plus the feed that syncs to channels like Google Merchant Center. On Shopify, this is what flows into Shopify Catalog and out to the AI channels. Gaps here are the most common reason a good product never gets mentioned.
Your product page
Assistants still read the page itself. Clear headings, a description that answers real buying questions and structured data markup all help. A description that's mostly mood and adjectives gives an assistant almost nothing to quote.
Your policies and support content
Shipping times, return windows and sizing help often decide a recommendation. If a shopper asks for something that arrives before the 20th and your shipping page is vague, you're out.
What other people say about you
Reviews on your site and mentions around the web help an assistant judge whether you're a safe pick. A product with real, specific reviews is easier to recommend than one with five stars and no words.
The fixes that matter most
You don't need a new platform for this. Most stores already have the pieces, they just haven't been filled in properly. Here's where we'd start, in order.
- Rewrite product titles to say what the thing is. "The Everyday" tells a shopper nothing. "The Everyday, a waterproof leather backpack for 15 inch laptops" tells an assistant exactly when to suggest it. Keep your brand name for the product, then add the plain description.
- Fill in every attribute your platform offers. Material, size, weight, color, compatibility, care, age range and GTIN or barcode where you have one. Empty fields are invisible to an assistant.
- Answer buying questions in the description. Think about what people ask before they buy: will it fit, how long does it last, how does it compare with the cheaper one, is it good for a gift. Answer those in plain sentences.
- Move key facts out of images. If your size chart, ingredients or specs only live in a graphic, an assistant can't read them. Put them in the text too.
- Keep price and stock accurate. Assistants lean on live data. A product that shows in stock but isn't, or a sale price that's expired, makes both the shopper and the tool less likely to trust you.
- Make shipping and returns easy to find and specific. "Ships in 1 to 2 business days, free returns within 30 days" is something an assistant can repeat. "Fast shipping" gives it nothing to work with.
- Ask for reviews that mention details. Prompt customers to say how they used the product, what size they bought and who it was for. That detail is what makes a review useful to the next shopper and to the assistant reading it.
None of this is new advice for good ecommerce. The difference is that gaps show up faster with AI shopping, because there's no beautiful photo to cover for missing information.
Where brand still makes the difference
There's a reasonable worry here. If every store fills in the same attributes, doesn't every product start to look the same?
Not if you're clear about who you're for. Assistants match products to specific requests, so a store with specific positioning matches more of them. A candle brand that says it's "for people who want their apartment to smell like a cabin in the woods" will match more real questions than one that calls itself "premium home fragrance".
So the work is the same work that sells to people. Know who your customer is, say it plainly and back it up with facts. A store that's consistent across product pages, collections and policies gives an assistant a clear story to tell. A store where every product was written by a different person in a different year doesn't.
If you've read our post on why ranking on Google doesn't mean AI recommends you, this will sound familiar. That one's about local service businesses, but the lesson carries over. AI tools recommend what they understand, and they understand what you've written down clearly.
How to check where you stand
You can test this today in about fifteen minutes.
- Open ChatGPT and ask for a product in your category the way a customer would, including a budget and a use case.
- Try three or four versions of the question, from broad to specific.
- Note who gets recommended, and look at their product pages next to yours.
- Check your own store for the gaps above, starting with your top ten sellers.
You'll usually spot the difference quickly. The products that keep coming up have clear titles, complete details and specific reviews. That's a gap you can close, and it'll help your conversion rate with human shoppers too, as we covered in why your Shopify store isn't converting.
What to do next
Start with your best sellers before Black Friday. Rewrite the titles, fill in the attributes, pull specs out of images and tighten your shipping and returns copy. Then run the same questions again in a few weeks and see what's changed.
If you'd like help, this is the kind of work our ecommerce team does every week: product pages, catalog structure and the brand positioning that ties them together, built for human shoppers and the AI tools they now ask first.
What would change for your store if every product page answered the questions your customers are already asking?