What a store needs from this check
An online store wants to know two things from an AI visibility check. Is the store named when someone asks where to buy? And are its products named when someone asks which one to pick? Those are different tests, with different fixes, and a store can pass one and fail the other.
That makes a store check different from a local business check, where place is the heart of the question, and different from a service business check, where the buyer wants a person. Here the buyer is shopping. They have a product type in mind, a use, a budget, and often a doubt about whether to spend more.
This page adapts the general method. The routine for asking and recording is covered in our general routine for running a check, and we will not repeat all of it. What changes for a store is the question set, the sources you read, and what you record about products.
Some people call the work of improving this GEO, short for generative engine optimization, or AEO, short for answer engine optimization. The labels do not matter much. The check is the same.
Test where-to-buy questions and which-one-to-pick questions separately. Record which products and which stores are named, and read the shopping sources behind them.
The question types for a store
Online shoppers ask assistants in a handful of recurring ways. Build your set from these, in your category's own words.
Where to buy. "Where can I buy X online." "Best store for X." These test whether you are named as a seller, and they often return lists of stores.
Best of. "Best X for Y." These return products, sometimes with the store named beside them.
Comparisons. "X versus Y for a small apartment." The buyer already has two options and wants help choosing.
Problem to product. "My dog chews everything, what should I buy." The buyer does not know the product name. This is where a store's own educational pages may matter.
Gifts. "Gift ideas for a new gardener under a set price." Common in seasonal categories.
Constraint questions. "Without a particular ingredient." "Fits a small space." "Made in the country." A single constraint can reshuffle the names completely.
A first set of twelve to fifteen questions can cover all six. Use the words your customers use in your support emails and reviews, not the words on your product pages. How to write and freeze the set is covered in the question set guide.
Map your questions to your products
A check that asks "best store for pet supplies" when you only sell reptile supplies will teach you nothing. Before you write the set, make a short map.
- Your top five products or collections. The things that bring in most of your revenue, or that you most want to be known for.
- For each, the question a buyer would ask before buying it. Write one where-to-buy, one best-of, and where it fits one comparison.
- The rival products buyers weigh against yours. Not the competing stores, the competing products. You will see them in answers.
This turns a vague wish ("be visible in AI") into a tidy, testable list. Five products with three questions each is fifteen questions, which is plenty.
If you are on Shopify, the structure of your store helps here, since collections and product pages are natural places for these answers to live. Our guide to Shopify and ChatGPT covers how the platform's pages feed what assistants can read, and the Shopify SEO checklist covers the basics underneath. Neither belongs in the check itself, but both help once you know what the check found.
Running the store check
Use a private window for each assistant and ask each question once, logged out. ChatGPT, Perplexity, Claude and Gemini are the four to start with. If you also want Google's AI answers, see how to check Google AI Mode.
A few things are specific to stores.
- Shopping answers differ. Some assistants show product cards, prices or shopping results beside a written answer. Others write only text. Note which kind you got, because the sources behind a product card may be different from those behind a sentence.
- Prices and stock change. An assistant may quote a price or say something is available. Note it, and check whether it is true. If it is wrong, the source of that fact is worth finding.
- Region matters. Shipping and availability depend on where the buyer is. Run from the country you sell to, or name it in the question.
- Watch for marketplaces. Large marketplaces and big retailers often appear in where-to-buy answers. Note them. They are part of the competitive picture, though not necessarily one you can compete with directly on the same ground.
Do not rephrase and re-ask until you get a good answer. The result you want is the first honest one, written down, in the same form you will use next month.
What to record for each answer
The recording matters more for a store than for a service business, because there are more moving parts per answer.
- Was your store named? As a seller, a brand or a source.
- Were your products named? Which ones, and with what description.
- Was a link shown to your store, and to which page? A link to a product page, a collection, your home page or a blog article tells you what the assistant thinks you are good for.
- Which other stores and brands were named, in what order.
- Which sources were shown or implied. Review sites, comparison articles, forum threads, retailer pages.
- What reason was given. "Best value," "most durable," "wide selection." Copy the wording.
The layout for keeping all this is described in an AI visibility tracking spreadsheet. A store with fifteen questions across four assistants is sixty rows a month, which is manageable but only if the columns are consistent.
One line of caution. It is easy to get absorbed in whether the answer names you and forget to write down who else was named. The competitor list is often the more useful half of the notes.
The sources behind store answers
When assistants answer shopping questions, the pages they rely on tend to come from a few families. Learn to recognize them in the source links.
Review and comparison articles. Independent writers testing or ranking products. If a few of these dominate your category, being included in them is a large lever.
Discussion threads. Forum and community posts where people compare what they bought. Our page on Reddit and AI search explains why these are weighted the way they are.
Retailer and marketplace pages. Product listings with ratings and reviews.
Brand and manufacturer pages. The official description of a product.
Your own content. Buying guides, comparison pages, care guides and FAQs on your store.
For a deeper look at what assistants tend to cite across categories, what AI engines cite covers the pattern. For your own case, the practical question is simple. In the answers where you are missing, which of these families do the sources fall into, and which of them could include you?
If the answers cite buying guides, a good buying guide of your own may earn a place. If they cite comparison articles from outside, the work is to be a product worth including and to make that easy to see. A buyer's guide written for a specific question is often the clearest example.
Named as a store, or named as a product
A result that gets overlooked is the gap between being named as a place to buy and being named as the thing to buy.
Store named, products not. You are known as a seller in the category. Your products may not be the recommended ones. That points to product pages, comparison content and reviews.
Product named, store not. The assistant recommends your product but sends the buyer to a marketplace or a different seller. You are visible and not capturing the sale. Check which pages talk about your product and whether they link to you.
Both. The best case.
Neither. Start with the sources and the competitors.
Record which of these four applies to every question. It becomes your priority list. A store that is named as a seller in ten answers and as a recommended product in none has a different job ahead than a store with the reverse.
For the interpretation of what each means, see reading your AI visibility results, and for what to do, what to do after a check.
How often, and how long it takes
For a store, a monthly check is a good rhythm. Product catalogs and review counts change slowly, and AI answers are noisy enough that a weekly check with fifteen questions will mostly show you noise.
The exception is a launch or a promotion. If you introduce a new product line, run the relevant questions a few weeks later, and again at two months. New pages take time to be read, and rarely show up in answers within days.
Time-wise, a first run of fifteen questions across four assistants is about an hour, mostly typing and noting. Later runs are faster. If that is more than you want, cut to eight questions and keep them fixed.
Repeat a few of the key questions two or three times in the same week now and then, to see how steady the answers are. Our page on why AI answers change explains what to expect, and how many repeats count as a fair reading.
If you want a single number to follow, "named in X of 60 answers" is honest and simple. Our page on the AI visibility score goes into what any single number can hide.
Mistakes stores make with this check
A few errors come up often enough to name.
Testing only branded questions. Asking "is [your store] any good" tells you what the assistant knows about you after you have named yourself. It is not visibility. Leave your name out.
Using product page titles as questions. Your page title is written for a search box. Real buyers say it differently.
Ignoring the competitor products. Buyers choose between products, so the rival product list is part of the result.
Treating one assistant as the whole picture. ChatGPT, Perplexity, Claude and Gemini draw on different material. A result in one is not a result in all.
Changing the wording each month. That breaks the comparison.
Panicking over one bad answer. Assistants are variable. Look at the pattern across questions.
Keep your name out of the questions, test both where-to-buy and which-to-pick, write down who else was named, and run the same set monthly.
A worked example: a small tea shop
Take an imaginary online tea shop. The owner picks her top product, a loose-leaf sleep blend, and writes three questions about it. "Where can I buy a caffeine-free tea blend that helps me wind down." "Best herbal tea for sleep for someone who hates chamomile." "Loose-leaf versus tea bags for a beginner."
In the answers, the first question returns a handful of stores and two big marketplaces. Hers is absent. The second names several products and links to a comparison article and a forum thread, and her blend is in neither. The third answers in general terms, with no store named.
Her notes show a clear pattern. Where to buy is dominated by larger sellers. Which to pick is shaped by the article and the thread. Her store shows up nowhere. She has no buying guide, her blend page is a short description, and her reviews sit in an app that does not show on the page.
The to-do list is concrete. A clear guide to sleep teas that says plainly what her blend contains and who it suits. A comparison page on loose-leaf versus bags. Reviews placed on the product page itself. And a plan to be included in the kind of roundup article the assistants kept citing.
None of that guarantees a result, and she will repeat the check next month to see. But she now knows what to build and why.
What RunOctopus does for a store, and what it does not
RunOctopus does not include a free tool that asks the assistants about your store. The manual check on this page is how to do that for free.
What it does is the build. Ollie, our AI, researches your business, then writes and publishes pages straight onto your own website, with a team that plans, writes, reviews and publishes alongside it. It works with Shopify, WordPress, WooCommerce and Wix. Nothing to install, upload or schedule, and you do not approve each page. In the first week you get 8 long-form articles, 6 pages and a custom interactive tool built for your niche, then 175 or more pieces of content a month, written, published and linked.
For a store, that is the buying guides, comparison pages and answers to the questions your check turned up. Paying customers also get reports on where ChatGPT, Perplexity, Claude and Gemini name their business.
All Angles Creatures, a reptile supply store and the place RunOctopus started, went from 0 to 4,500 or more Google clicks a month in five months, per Google Search Console for the 28 days to September 29, 2026.
The free site read reads your store address and shows the plan. No account, no call, and no check of AI answers.