AI visibility for online stores
Buyers increasingly start with a question rather than a search box: where to buy, which brand is better, who delivers to my area. The answer names a handful of stores. This page explains what decides whether yours is among them, and what usually prevents it in a shop.
A buying question is not a search string
In a search box a buyer types a product name. To a model they put a question with conditions: for what use, in what price range, with what delivery or warranty. The answer therefore does not pick between pages but between vendors it has enough description about to check the conditions in the question.
That leads to an uncomfortable conclusion for many stores: a catalogue of names, images and prices is nearly empty for a model. What decides are sentences about who the product is for, what is in the box, when and where you deliver and on what terms it can be returned.
What most often stops a crawler in a store
- A catalogue that renders only after scripts run. The crawler receives a page shell with no products, although a buyer sees a full store in the browser.
- Default robots.txt rules from a template or plugin that close off categories and product pages along with the cart and filters.
- A product description copied from the supplier and identical at ten retailers. It says nothing about why a buyer should buy from you.
- Delivery, return and warranty terms only in fine print or behind a form. What is not in the page text cannot be used by a model answering a conditional question.
- No mention outside the store itself: not in comparisons, not in buyer reviews on independent sites, not in write-ups about the brand you sell.
Structured data: what the score counts and what it does not
Our score reads seven JSON-LD types: Organization, Corporation, LocalBusiness, ProfessionalService, Service, FAQPage and SoftwareApplication. Product and Offer are not among them. That does not mean they are pointless; it only means they earn no points in this score, and we will not write it otherwise.
What earns a store points here is a record about the business itself: who you are, where you are, what you sell and to whom. That record is what online stores most often lack, because attention stopped at the products.
Invented product ratings in the markup are a separate story and a real risk. An average-rating field without genuine reviews on the page is grounds for a manual search penalty, which is why it appears nowhere in our markup.
What to measure first
Start with one measurement of the domain. The free one asks eight questions on three models and runs a technical check; that is enough to answer whether any of them mentions you and whether the technical side blocks you. The paid report asks one hundred and twenty questions, each three times on each model, and shows who is named instead of you.
The order of fixes mirrors the score: access first, then readable content, then the record about the business, and only then content that others cite. Working in that order saves effort, because without the first step the other three are visible to nobody.
Frequently asked questions
Is every product in the catalogue measured?
No. The measurement asks the kind of questions a buyer asks and looks at whether your store is named in the answer. The unit is the domain, not an individual product page; a catalogue of a thousand products does not become a thousand measurements.
I sell brands that ten other retailers also sell. Is a measurement meaningful at all?
That is exactly where it is most useful. When the product is identical, the answer turns on what differs: delivery, the price with its terms, warranty, advice, service. The measurement shows whether that is written in the text or stays only in your head.
Should I block AI crawlers so they do not take my product descriptions?
That is your decision and it can be entirely legitimate. What matters is knowing the consequence: a closed door also means the model cannot name you in a buying question. The mistake is not a deliberate block but a block you do not know about.
How quickly does a fix show up in the answers?
The technical part of the score changes at the next measurement, because it is measured directly on your site. Presence in the answers changes more slowly and unevenly, because it depends on when a provider refreshes a model and on what is written about you elsewhere. That is why we track the delta between runs rather than a single number.
Measure your store
Eight questions on three models, a technical check and a score from 0 to 100. No card required.
Check AI visibility