AI ReadinessWhat we check

28 checks · Shopify · free

Find out what AI shopping agents can actually read on your store.

Shoppers increasingly ask an assistant before they open a browser tab. This scan fetches your public pages the way those agents do — robots.txt, structured data, your MCP endpoint, your policies — and shows you, with evidence, exactly where you are invisible to them.

Free, no login, no changes to your store. Results in about a minute.

One more step

Enter your email and we will run the scan on and send you the report.

Why do we ask for this?

Each scan makes real requests to a real storefront, so we keep them attributable — and it is how we send you the report. We never sell your address.

A score you can argue with

Every finding cites the URL, the snippet and the count we saw. Nothing is asserted without evidence.

An AI crawler matrix

14 named bots — OpenAI, Anthropic, Perplexity, Google, Apple, Meta — tested against your real paths.

A ranked work order

Fixes sorted by impact and effort, so the first three are the ones actually worth doing.

Honest gaps

Anything we were prevented from testing is marked untested, not counted against you.

What we check

Four categories, weighted by how much each one affects whether an agent can find, understand and recommend your products.

Basic

25%
  • robots.txt reachable and parseable
  • 14 AI crawlers tested against product, collection and content paths
  • llms.txt present, valid, and at the right path
  • Storefront MCP server responds and answers policy questions
  • UCP merchant profile published
  • Catalog completeness against Shopify Catalog requirements

Site-wide

20%
  • Shipping, refund, privacy and terms policies
  • Whether policy content survives without JavaScript
  • About page with facts an agent can repeat
  • Contact options a machine can read
  • FAQ coverage and FAQPage structured data

Product pages

35%
  • Product and Offer JSON-LD completeness
  • GTIN / barcode identifiers, per variant
  • Specifications in descriptions, not just adjectives
  • Review ratings as structured data
  • Shipping and returns on the page itself
  • Image alt text across the catalog
  • Price and title present without JavaScript

Long-tail

20%
  • Comparison and "best for" content
  • Use-case and compatibility pages
  • Buying guides and sizing help
  • Collection descriptions

Being on Shopify is not the same as being readable.

Shopify has done a lot of this for you. Your store probably already exposes an MCP endpoint that agents can query, a machine-readable catalog, and a robots.txt that welcomes AI crawlers by default.

What breaks is everything layered on top: a theme that drops structured data, a policy page that only renders after JavaScript, a barcode field nobody filled in, a robots.txt edited in 2023 to block GPTBot and never revisited. Each is small. Together they decide whether an agent can quote you.

This scan finds them and tells you which ones are worth your afternoon.

Questions

Do you change anything on my store?
No. Every check is a read of a public URL — the same pages any shopper or crawler can already fetch. We never sign in, and we make no writes of any kind.
How many requests does a scan make?
Between 25 and 40, capped at 80, spaced at least 200 milliseconds apart with at most four in flight. If your store starts refusing us, we stop asking rather than pushing through.
What if a check cannot be run?
It is reported as untested and left out of the score entirely. "We could not measure this" and "this is broken" are different statements, and we will never show you the second when we mean the first. If too much of the scan is blocked, we withhold the score rather than publish a misleading one.
Does this work on headless or Hydrogen stores?
Yes. Where the usual Shopify routes are missing we fall back to your sitemap and navigation. You will often score lower, because content that only exists after JavaScript runs is invisible to most AI crawlers — which is the finding, not a limitation of the scan.
Why does structured data matter more than my copy?
An agent comparing three merchants needs your price, availability, brand and identifiers as unambiguous fields. Prose it has to interpret is a tiebreaker at best. Stores that publish clean structured data get quoted; stores that do not get skipped.
Is my scan public?
The report has a shareable link, so treat it as public. Everything in it was already publicly readable on your storefront.