An agent cannot buy from a catalog it cannot read. We scored 245 Shopify storefronts on readability, and six of them reached the second highest band. Zero reached the highest.
- Readability is the precondition for agentic commerce. Everything else is downstream of it.
- Of 245 Shopify storefronts, 6 reached Established and 0 reached Leading.
- The failures are boring and fixable: missing attributes, unstated availability, claims buried in prose.
What does agent readiness actually require?
Four things in sequence, and a failure at any one of them ends the transaction before the next begins.
What did we measure, and what did we not?
We measured the first three. Our score covers whether a site can be reached, whether its meaning can be extracted, and whether the data an agent would rely on is present and consistent. That is the readability precondition.
We did not run purchase attempts against 245 live stores, because that would mean placing real orders in other people’s shops. So read this as a measure of whether an agent could reasonably try, not as a count of completed transactions. Being clear about that boundary matters more than a bigger claim would.
How ready is the Shopify population?
Barely. The distribution across 245 storefronts, scored on the same engine and methodology version, breaks down like this.
| Band | Stores | Share | What it means for an agent |
|---|---|---|---|
| Starting Point | 68 | 28% | Substantial gaps. An agent would struggle to describe it |
| Developing | 171 | 70% | Readable in parts. Comparisons would be unreliable |
| Established | 6 | 2% | An agent could work with this |
| Leading | 0 | 0% | Nobody |
What stops a catalog being agent ready?
The same handful of things, over and over, and none of them are exotic.
- Attributes that exist only in an image, so size, material and fit are invisible.
- Availability that renders after a script runs, so stock status reads as unknown.
- Claims made in prose rather than fields, which is the pattern in claims belong in fields.
- Structured data that disagrees with the visible page, which is worse than having none.
- Identifiers missing, so the same product cannot be matched across sources.
Why is the empty top band the important number?
Because it means the standard has not been set. In a category where the leader is at 69 out of 100, being genuinely good is available to anybody willing to do unglamorous work for a few weeks.
That window closes as the population improves. The stores that fix this while the median sits at 46 will be the ones agents have been reading for a year by the time it matters.
Method
Cohort: 245 public Shopify storefronts scored at methodology version aivs-1.0-aeo20260620-full-evid20260621, read live from our benchmark on the publication date. Full cross platform cohort for comparison: 613 sites, median 47.2.
The full population analysis is in the Shopify legibility study.
Common questions
Can AI agents really complete purchases now?
Increasingly, through several mechanisms that differ by platform and checkout. The direction is consistent even where the implementations are not.
Does this mean 98% of stores cannot be bought from by an agent?
No. It means 98% fall short of the readability level where an agent could work with the catalog confidently. Many transactions will still complete. Fewer will start.
What is the single highest value fix?
Completing product attributes as structured data. It is the input to every comparison an agent makes, and it is usually the emptiest field set on the store.
How do I find out where my store sits?
Run the free scan. It returns the same 0 to 100 score used in this study, so your number is directly comparable to the bands above.
Related reading
Run a free scan and get a 0 to 100 score with a fix list in priority order. Run the free scan.
Your buyers are already asking AI. This is how you make your website readable to the assistants they ask.
Everything in this article is measurable on a live storefront. The Full Assessment and Roadmap reads your website the way AI crawlers receive it and hands you every fix in plain English, in the order we would make them.