Show what your solution changes for AI shopping agents

You sell product data enrichment, feeds, structured data, a UCP plugin, or a checkout, payment or delivery widget. We put the same products into two reference stores, one before and one after your solution, and run the same shopping requests through 8 AI models. You get the measured difference, with a confidence interval and the recording of every run.

You pay after the report. Results stay private unless you agree to publish them in full.

Everything happens in writing: scope, main measure and price are agreed by email before the test. No calls needed.

Ask for a test How we test

Who it is for

Product data enrichment: Richer titles, descriptions and attributes. Does an agent pick a product that fits the request more often?
Feeds and structured data: Product feeds, JSON-LD, GTINs and variants. Does the agent read the right price, availability and variant?
UCP and agent plugins: A machine-readable way into the store for agents. Do agents use it, and does it change what they buy?
Checkout, payment and delivery widgets: Express checkout, payment choice, pickup-point pickers. Does the agent still reach checkout, within budget?

How the test works

You write to us through the form below. Within 1 business day we reply by email with a proposed scope: which products, which shopping requests, and the main measure. The main measure is written down before anything runs.
You send your product data, or access to your solution. We load the same products into two reference stores: A as the products usually are, B with your solution.
We run the same shopping requests through 8 AI models in both stores, 2 or 3 times each: 160 to 480 recorded runs, depending on the scope.
Within 7 days of receiving your data you get the report and every recording. You pay after you receive it.

What we measure in every run

Did the agent pick a product that fits the shopping request?
Did it reach the checkout?
Did it stay within the shopper’s budget?
How many steps did it need?

What you get, and the price

What you get

The difference between store A and store B on the main measure, with a confidence interval.
The same comparison for the other measures: product fit, checkout reached, budget kept, steps used.
The recording of every run, in both stores and with every model.
The cases where your data changed the agent’s choice, side by side.

Price and terms

One solution, one scope agreed in writing.

$1,500 per test, paid after you receive the report.
Results within 7 days of receiving your data.
Retest after a change to your solution: $490.
Scope, main measure and price are agreed by email before the test.

No calls, no subscription. If the scope does not fit, we say so before you send anything.

Independence rules

You pay for the test, not for the result. The price is the same whatever the result shows.
The same method for every vendor. How we test: method.
The main measure is written down before the test and does not change after it.
Results are published only with your consent and only in full. If you cite a result, you link to the full result.
Everything is agreed in writing, by email.
The agent never pays, never places orders and never types personal or payment data. Details: our agent.

Limits

Our 20-product reference store is a demonstration. In a paid test we load the products you choose.
We test with our own agent and 8 AI models. Consumer agents wrap models in their own tools and rules, so their results can differ.
A result describes our stores and these shopping requests. It does not promise the same effect in every store or with every agent.

Ask for a test

A few sentences are enough: what your solution does, which products you would like to test with, and what you want to show. We reply by email within 1 business day.