Skip to content
for online retail

a shopper asks what to buy. is it your product?

Product research now starts with a question rather than a query, and the answer names two or three products. That answer is a shelf, and either you are on it or you are not.

No account, no card. Usually under ten minutes.

You sell products online, shoppers research before they buy, and you want to know whether an assistant puts you in the answer or sends the sale to a marketplace listing.

the questions

what your buyer actually types

These are the shape of the questions we ask on your behalf. You never write them: a set is drawn from a shared library for your category, and on the higher plans a set is written specifically for you.

  • What is the best running shoe for flat feet?

  • Which coffee subscription is worth it?

  • What should I buy for a first-time camper?

  • Best budget standing desk under 400 dollars

  • Which skincare brands are actually fragrance free?

Each question is asked repeatedly rather than once. Two identical runs minutes apart agree on only 36% of the brands they name, so a single answer is one draw and not a measurement.

why it matters here

what is actually at stake

The answer is a shelf with three spaces on it

A results page has ten links and a shopper who scrolls. An assistant answer has two or three products and a reason for each. The shelf got shorter, which means the cost of being off it went up, and nothing about your product page tells you whether you are on it.

Marketplaces are the default answer

Ask for a product recommendation and the sources cited are frequently a marketplace listing, a review roundup or a comparison site rather than the maker. If the assistant names your category and then cites somebody who resells you, the demand exists and the visit does not reach you.

Product claims are repeated, correctly or not

An assistant summarising your product will state its specifications, its price band and who it is for. Those come from whatever it read. Measuring the answer is also how you find out it is telling shoppers something about your product that stopped being true a year ago.

questions

the things people ask us before they start

How do I know if AI recommends my products?
Ask the buying question a shopper would ask, repeatedly, and record the answers. unbound geo runs each question ten times on the consumer surface and three through the provider API, because two identical runs agree on only 36% of the brands they name, so one answer is a draw rather than a measurement. You get how often your product was named, which competitors were, and the pages cited each time.
Why do AI assistants recommend marketplace listings instead of my store?
Because marketplace and review pages are frequently what the assistant read. The cited sources are visible in every answer we store, so you can see whether the answer about your category rests on a reseller listing, a review roundup or your own product page — and that tells you where the gap is rather than leaving you to guess.
Does this replace shopping feed or marketplace tools?
No, and it does not try to. Those manage listings and ads inside a marketplace. This measures what a general-purpose assistant says when somebody asks what to buy, which is a different surface with different sources, and neither one tells you about the other.
where to start

The scan is free and needs no account. Most companies like yours start on Starter, at 49 USD a month, and the per-call provider costs behind every price are published.

See pricing
what it does not do

This measures recommendations and the sources behind them. It does not track your listings on any marketplace, and it does not read your sales data, so it cannot attribute a sale to an answer. What it gives you is what was said and what it was based on.

free scan

are you in the answer?

Five real questions against a live answer engine with web search on, each asked ten times. That is fifty real calls, so the report lands in under ten minutes rather than while you wait. You see every question before anything runs, and you get the answer text, who got named and the sources the engine read, not a score.

your site/what you sell/where your buyers are/the questions