For ecommerce

Getting your products recommended by AI

Nobody asks an assistant for the best online store. They ask for the best product of a type, and that one difference changes almost everything about where the work goes.

The question is product-shaped, not brand-shaped

A software buyer asks “what is the best help desk tool” and the answer names companies. A shopper asks “best waterproof hiking boots under $150” and the answer names products, usually pulled from a roundup, a marketplace listing or a forum thread where real people argued about it.

This is why most GEO advice does not transfer. A tool that tracks your brand name will happily report that nothing mentions you, which is true and useless, while every answer in your category names three products you compete with. The thing to track is the question a shopper actually types.

Where the recommendation is decided

Almost never on your product page. Your page is one vendor making a claim about itself, competing for retrieval against a review that compares ten options and a thread where twenty people said which one lasted.

When we measured 3 software categories on 26 July 2026, reading every one of the 131 sources behind 18 answers, the deciding pages were third-party in all three, and community sources outnumbered review directories heavily. We measured software rather than physical products, so treat that as a strong hint rather than a law for your category. The hint points somewhere useful: roundups and forums over your own catalogue.

What to do, in order

1. 1

   Write the ten questions a shopper actually asks

   Not keywords. Full questions with the qualifiers people really use: price ceilings, use cases, comparisons against the obvious alternative. This list is the whole measurement, so it is worth an hour.

2. 2

   Find out which pages answer them today

   Run the questions and read the sources. You will usually find the same handful of roundups and communities across most of them, and that list is your target rather than a keyword report.

3. 3

   Get into those pages

   Pitch the roundup that already ranks. Answer honestly in the thread where people are comparing. Make sure marketplace listings carry the specs a summariser needs. None of this is glamorous and all of it beats another round of on-page tweaks.

4. 4

   Make your own page easy to summarise correctly

   This is fourth for a reason. Clear specs, real answers to buyer questions, honest reviews. It does not get you cited, it makes the citation accurate when it comes, which is worth having and is not the lever.

What we will not claim

That we can get your product into an assistant’s answer. Nobody can promise that, and the measurement is the honest part of this: we can tell you which pages decide your category and whether your position in them changed. What you do with that list is the actual work.

Questions ecommerce teams ask

What is the best AI visibility tool for an ecommerce store?

The one that lets you ask product-shaped questions rather than brand-shaped ones, and shows you the sources behind the answers. Shoppers do not ask an assistant for the best store, they ask for the best product of a type, so a tool that only tracks your brand name will report almost nothing while you are being left out of every answer that matters.

How do I get my products recommended by ChatGPT?

By being in the pages an assistant retrieves when someone asks for that kind of product: roundups, best-of lists, marketplace listings and forum threads. Your own product page matters much less than most guides suggest, because it is one vendor page competing with a review that names ten options.

Does optimising my Shopify product pages help?

It helps once someone is already looking at you, and it does not get you into the shortlist. Clear specs, real answers to buyer questions and honest reviews on the page make you easier to summarise correctly when you are cited. They do not put you inside somebody else's roundup, which is where the recommendation is decided.

Do reviews and forums matter more for ecommerce?

In our measurement of software categories, community sources accounted for 21 of 131 cited sources while review directories accounted for 1. We measured software, not physical products, so treat that as a strong hint rather than a finding for your category, and check your own: it takes one free query.

How many questions should I track?

Fewer than you think, chosen well. Ten genuine buying questions in your category, phrased the way a shopper phrases them, beat a hundred keyword variations. The point is to be able to compare the same questions over time, and a set you can read in one screen is a set you will actually maintain.

Our published measurements cover software categories, not physical products. Every figure above that comes from them is marked as a hint rather than a finding for your category, because that is what it is.