Do shoppers use ChatGPT to find products?

How many shoppers start product research in an AI assistant?

The numbers from several large surveys point the same way.

PYMNTS Intelligence found that nearly 61 million US consumers now begin their online retail research with AI assistants, based on an August survey of 2,191 American adults. Omnisend found that 63% of Americans used generative AI for shopping-related tasks in the past six months, and that ChatGPT leads that shopping use.

The pattern is not limited to the United States. NielsenIQ found that 55% of global consumers report some level of AI influence on product discovery in consumer goods, in a survey of more than 21,000 consumers across 31 countries. Clearpay found that 46% of UK consumers have already used AI tools to research products or compare shopping options, rising to 59% among Millennials.

Do shoppers buy where they research?

Most of them do not, and that changes how a store should read its own numbers.

CivicScience found 61% of AI shoppers researching on a platform and then buying elsewhere. In practice, a shopper asks an assistant which running shoes suit flat feet, reads three names, and later types one of those names into a search bar or goes straight to a marketplace.

The assistant shaped the choice, but the store's analytics may show a search for the store's name or a direct visit instead. If your store was not among the three names, nothing in your reports shows the sale you never had a chance to win.

How often are stores missing from AI answers?

In one recent test, most brands were not named when the shopper described a need instead of a brand.

Digital Commerce tested 25 direct to consumer and e-commerce brands in 14 categories across ChatGPT, Claude and Perplexity. Digital Commerce found that 80% of them, 20 of the 25, were invisible at the problem-aware stage, when the shopper describes a problem without naming any brand.

The same study found a stranger result. For 12% of the brands Digital Commerce tested, 3 of the 25, the brand's own content was cited as a source while a competitor was recommended instead. Digital Commerce calls these ghost rankings: the assistant read the store's page and still sent the shopper somewhere else.

That is the core risk for a store. In the Digital Commerce test, being cited as a source did not always mean being recommended, and the only way to know which case applies to you is to look at the answers themselves.

How can a store check what ChatGPT says about it?

You can run a first check yourself in one sitting, as long as you keep the method fixed.

1. Step 1**Write the questions your shoppers ask.** Use their words, for example "best waterproof hiking backpack for long trips" or "gift for a coffee lover who has everything". Keep most of them free of your brand name.
2. Step 2**Freeze the list.** Save the questions before you start, so a later check compares the same thing.
3. Step 3**Ask every question in several assistants.** Use ChatGPT, Claude, Gemini, Grok and Perplexity, with a fresh conversation for each question.
4. Step 4**Record five things per answer.** Is your store named, is it recommended, how high does it appear, is a source cited, and is the description of your products correct.
5. Step 5**Record the same for your two or three closest competitors.** Your result only means something next to theirs.
6. Step 6**Mark the ghost rankings.** Note every answer that cites your own page while it recommends someone else, because those are the clearest gaps to study first.

What should a store look at first?

Start with the answers where the gap is clearest, then work outward.

1. Step 1**Answers that cite you but recommend a competitor.** The assistant already uses your page as a source, so compare what your page says with what the competitor's page says for the same need.
2. Step 2**Wrong facts about your products.** Outdated prices, discontinued items or a wrong category in an answer should be corrected at the source pages the assistant cites.
3. Step 3**Needs where no store in your category is named.** These questions are open, and clear product pages that answer the exact need are where to start.
4. Step 4**Assistants that never name you.** Each assistant can answer differently, so a store can be visible in one and missing in another.

Keep the same question list for every later check. With the same questions each time, you can see whether the answers changed after your work.

How does EntityRise measure this for a store?

EntityRise is an AI visibility audit for agencies and companies that shows how ChatGPT, Claude, Gemini, Grok and Perplexity describe and recommend a company and what to improve first.

Each audit uses 20 buyer questions, 17 of them discovery questions without the brand name and 3 brand-aware, across five AI assistants, for up to 100 answers kept with their sources. Before anything is measured, the business profile, the buyer topics and the question set are confirmed and then locked. The AI Visibility Index weighs recommendation at 30%, mention at 20%, prominence at 20%, citation support at 15% and description accuracy at 15%.

The result is a PDF report of around 40 pages with the visibility analysis assistant by assistant, the competitor comparison and the ordered list of what to improve. You can start the audit yourself, and agencies can run it for their e-commerce clients under their own name. EntityRise measures and guides; the improvements happen on your own pages and in your own marketing work.