How to get your products recommended by AI

August 4, 2026
Shopify Tips
10 min read
an AI chat speech bubble spotlighting and pointing to one product card highlighted among several dimmed product cards, representing an AI assistant recommending a specific product

Getting your products recommended by AI is quickly becoming its own channel, and most stores have no idea whether they are in it or not. A shopper opens ChatGPT and types “best waterproof hiking boots for wide feet under 150 dollars”. The assistant answers with three or four specific products and where to buy them. If your boots are not in that answer, you did not lose the sale on your product page. You lost it before the shopper ever became a visitor.

This is the uncomfortable part: you cannot see it happening. There is no line in your analytics that says “ChatGPT recommended a competitor instead of you 400 times this week”. The channel is real, it is growing, and for now it is nearly invisible to the people it affects most. So the honest first question is not “how do I rank number one”, it is “am I in the answer at all, and if not, why not”.

We build Shopify tools, and we are building a new one, Octolift, specifically because this problem has no good answer yet. This guide is the playbook: how AI assistants decide what to recommend, the concrete levers you actually control, and a five-minute way to check where you stand today. No fluff, and no pretending anyone has this fully solved.

In this post

How do you get products recommended by AI?

You get recommended by making your products easy for an AI to read, verify, and trust. In practice that means clean structured data on every product page, a crawlable store that does not block AI bots, third-party signals like reviews and mentions the model can corroborate, and content that answers real buying questions directly. Being indexed is not the same as being recommended, and the gap between them is where the work lives.

That last point deserves emphasis, because it is the thing most guides get wrong. An AI can know your page exists and still recommend a competitor. So the goal is not visibility for its own sake. It is being the answer.

Why AI recommends some stores and skips others

Here is a finding that should change how you think about this. Search Engine Land’s Lily Ray analyzed 100 “best [category] software” queries in 2026 and found that in 224 of 323 citations, roughly 69 percent, Google cited a brand’s own page and then excluded that brand from its recommendations, often recommending competitors named in the brand’s own article. Being cited is not being recommended. A page can rank, get quoted, and still send the customer somewhere else.

So what tips an assistant toward recommending you? A handful of levers show up again and again:

  • Machine-readable product data. Structured data (Product, Offer, Review schema) tells the model exactly what you sell, for how much, in stock or not, and how well reviewed. Ambiguity gets skipped. Our Shopify schema markup guide covers the fields that matter.
  • Corroboration from elsewhere. Assistants trust claims they can verify in more than one place. Reviews, mentions in roundups, and a consistent presence across the web all raise confidence. A store that only talks about itself is easy to leave out.
  • Direct, specific answers. Pages that answer the actual question (“is this waterproof”, “does it fit wide feet”) in plain sentences are far easier to quote than pages of marketing copy. This is the core of answer engine optimization.
  • Access. If you block AI crawlers in robots.txt, you opt out of the whole channel. We wrote about why you should not block AI bots for exactly this reason.
  • A machine-facing summary. An llms.txt file gives assistants a clean map of what your store is and sells. See our llms.txt setup guide.

None of these is a magic switch. Together they move you from “a store the AI has heard of” to “a store the AI is comfortable putting its name behind”.

Two problems hiding in one: training data vs. live retrieval

Why isn’t your brand in AI answers? Because there are two completely different reasons, and they have two completely different fixes. Most people conflate them and then fix the wrong one.

Training data. Some of what a model “knows” is baked into it from when it was trained. If your brand was barely mentioned online back then, it is barely in there now. You cannot edit training data, and the next training run is not on your schedule. This is the slow, reputation-shaped half, and the only fix is the long game of being mentioned more, everywhere, over time.

Live retrieval. The other half is real-time. When an assistant answers a shopping question, it often searches the live web and pulls in pages then and there. This half you can influence this week: schema, llms.txt, crawlability, direct-answer content, fresh reviews. A brand invisible in training data can still get recommended through retrieval if its live pages are clean and quotable. This is the winnable game, and it is where your effort pays back fastest.

The catch is that you cannot fix what you cannot see, and right now checking your live retrieval visibility means manually asking five different assistants your top queries and reading the answers by hand. That is exactly the gap we are building Octolift to close: it monitors whether AI assistants recommend your Shopify products and flags the AEO gaps keeping you out of their answers. It is launching soon, and the waitlist is open. Drop your email and store URL and you will be first to see how your store scores at launch.

Join the Octolift waitlist →

The checklist to get recommended

Concrete and in priority order. Everything here targets the live-retrieval half you actually control.

  1. Ship complete product schema. Product, Offer, and Review JSON-LD on every product page, with price, availability, and data populated. Generate and validate it with our schema generator.
  2. Unblock AI crawlers. Check robots.txt is not blocking GPTBot, ClaudeBot, PerplexityBot, or Google-Extended. Blocking them is opting out of the channel.
  3. Publish an llms.txt. A short machine-readable summary of what your store sells, so assistants map you correctly.
  4. Answer buying questions in plain text. On product and collection pages, answer the real questions (fit, materials, care, comparisons) in short, quotable sentences, plus FAQ schema.
  5. Build corroboration. Get reviews on your pages and mentions off them. One-sided self-description is the easiest thing for a model to discount.
  6. Keep it fresh. Update prices, stock, and content. Stale pages read as lower-confidence sources.

Not sure where your store stands on the basics? Our free AI readiness checker gives you a quick read, and the deeper mechanics live in our Shopify AI readiness and ChatGPT optimization guide.

Join the Octolift waitlist →

How to check whether AI recommends you right now

Before you change anything, get a baseline. It takes five minutes and it is sobering.

  1. Write down your top three buying queries the way a customer would type them (“best X for Y under Z dollars”).
  2. Ask each one in ChatGPT, Perplexity, Gemini, and Copilot. Use a fresh session so your history does not skew it.
  3. Note, for each: are you recommended, merely mentioned, or absent? Which competitors show up? Which sources did the assistant cite?
  4. Repeat monthly. AI answers shift fast, so a single snapshot is a start, not a verdict.

Doing this by hand across four assistants and a dozen queries every month is a chore, and it is the chore Octolift automates. But even the manual version is worth it once, because most merchants have never actually looked, and you cannot prioritize a fix until you have seen the gap.

Frequently asked questions

How do I get my products recommended by ChatGPT?

Make your product pages easy for it to read and trust: complete Product and Review schema, an llms.txt summary, AI crawlers unblocked, direct answers to buying questions, and third-party reviews and mentions it can corroborate. Then check monthly whether you actually appear for your top buying queries.

Why isn’t my brand showing up in AI answers?

Usually one of two reasons. Either you are thin in the model’s training data (a slow, reputation problem), or your live pages are hard to retrieve and verify (a fast, fixable problem: schema, crawlability, clear content). The live-retrieval half is where to start, because you can change it this week.

Is being cited by AI the same as being recommended?

No, and this is the trap. Analysis of “best software” queries found brands cited as a source but excluded from the recommendation about 69 percent of the time, with competitors recommended instead. Aim to be the answer, not just a footnote in it.

Does schema markup help LLM visibility?

Yes. Structured data removes ambiguity about what you sell, the price, availability, and ratings, which makes your product safer for an assistant to recommend. It is the single most concrete, actionable lever, which is why it tops the checklist.

How often should I check my AI visibility?

Monthly at least. AI answers change as models update and as the live web changes, so visibility is a moving target, not a one-time setup. A monthly check on your top buying queries catches drops before they cost you a season.

Do I need a tool, or can I do this manually?

You can start manually: ask your buying queries across the major assistants and log the results. That works for a first baseline. It gets tedious fast across multiple assistants, queries, and months, which is the repetitive tracking Octolift is built to automate.

The stores that win the AI channel are not the loudest. They are the easiest to read, verify, and trust. Fix the live-retrieval basics, then watch your top queries every month so you know the moment the answer changes.

Written by Ozkan Oz, founder at Craftshift, where we build Shopify apps and are building Octolift to help merchants get their products recommended by AI.

Join the Octolift waitlist →

http://craftshift.com

Co-Founder @ Craftshift

You may also like