Example
What an agent readiness report looks like
This is a full report, with nothing locked. It is what a merchant gets after a scan: the score, where the agent stopped, every finding with the change that fixes it, and the buy attempt step by step. Switch to the free view below to see what a free scan shows before you unlock it.
This store does not exist. The findings below are real failure modes from our stall library, arranged on a shop we made up so we could show you the format without publishing somebody else’s results. No number on this page is a measurement of anything.
How to read it
A report has four parts and they answer different questions. Taken in order they go from how bad is it, to where exactly, to what do I change, to who do I send this to. The same walkthrough is in help as reading your report.
The score, and what it is not
The score covers all four stages. A scan that did not reach checkout uses the last buy attempt from the week before and says when it was, and with none in that week there is no score, so the buy stage is never quietly scored as passing. It is a way to see movement between scans, not a grade. Two shops on the same number can have completely different problems, which is why nothing on this page asks you to act on it by itself.
The four stages
This is the useful part. The shop above reads well and cannot be bought from, which is the common case: discover and evaluate pass, and the walk stops at the size control. A shop that fails at discover has a different problem entirely, because nothing later in the list ever gets a chance to run.
What a finding contains
Each one names the stage it belongs to, how serious it is, what the agent saw, and the change that resolves it. The severity is not our opinion of your shop: it comes from the stall library, the same published list every scan is scored against, so the word means the same thing on every report we produce. That is also why a Critical needs two agreeing signals before we will print it, and why a report sometimes leaves a finding out.
Findings without a screenshot are not weaker. Some come from what the pages say rather than from what a browser showed: no llms.txt and no GTIN are read out of the data, and there is nothing to photograph.
The buy attempt, step by step
The walk is the part merchants forward. It lists what the agent did, what it found, how long each step took, and the screen it was looking at when it stopped. A developer reading it does not have to believe the score. They can open the same page and see the same thing.
What to do with it
Send it to whoever changes your theme. Most fixes here are small and none of them are about redesigning anything: a size control that is a real form field, a policy written as text, a popup that does not cover the basket. The report is written to be forwarded without translation.
A real scan works the same way and stops at the same place: before payment, with no order placed. What a scan does, and what it does not do sets out the whole boundary, and what we store and for how long covers the evidence a report keeps.
Run this on your own store, free, once. Or read why it matters first.