DIAGNOSTIC / 15 POINTS

Is your store ready for AI shoppers?

Assess catalog data, discovery, policies, authorization and control. This does not measure how “modern” your store looks; it identifies what an agent would need to understand and verify before recommending or facilitating a purchase.

READ THE WORKED EXAMPLE AND GUIDE ↓

Educational self-assessment. It is not a security, compliance or platform-compatibility certification.

WHAT TO DO WITH THE RESULT

Do not try to fix all 15 points at once.

01FIX COMMERCIAL TRUTH

Price, stock, variants, fulfillment and policies need to be correct before optimizing for agents.

02MAKE IT RETRIEVABLE

Structure information so people, search engines and AI systems can identify it without inferring the essentials.

03DEFINE ACTION BOUNDARIES

Before exposing actions, define authorization, confirmation, logs and payment handoff.

READING AND APPLICATION GUIDE

Turn each checkmark into evidence.

The score is a conversation starter, not a certification. Mark a point only when you can show the page, record or behavior that supports it. A documented limitation is more useful than an optimistic yes. Keep the date and owner of each observation so a later review can distinguish a real improvement from a changed assumption.

Worked example: a store scores 8 out of 15

This fictional store can verify four catalog points, three discovery points and one control point. Its total is 8/15. That number does not mean it is halfway to safe automated purchases: the weak control category can block action even when descriptions and structured data look strong.

Its first improvement is to reconcile variant stock with the checkout and define an explicit purchase review. Its second is to publish missing delivery restrictions. Repeating the scorecard after those changes requires checking the actual behavior, not simply changing the answers. A product feed or schema markup alone does not resolve those gaps.

Practical criteria for reviewing the example
Area	Evidence to collect	A useful next action
Catalog truth	A product and its variants compared with the authoritative inventory and policy pages.	Resolve contradictory units, price or stock before generating new descriptions.
Discovery	Readable product content, stable identifiers, canonical URLs and matching structured data.	Choose one product family and make its facts consistent across all surfaces.
Action and control	A purchase review that refreshes the offer and records explicit authorization.	Test a stock change, expired quote and rejected authorization.

An exercise for your project

Choose one real product, one variant and one delivery destination. Record the source of every claim a buyer needs.
Walk from discovery to purchase review without completing a charge. Compare the facts at each step and note contradictions.
Assign each gap an owner and a verification step. Repeat after the fix using the same product, plus a difficult variant.

Concepts worth distinguishing

Canonical URL: The preferred public address for a piece of content, used to distinguish its main version from duplicates.
Structured data: Machine-readable facts that must agree with visible content; they do not guarantee ranking or agent support.
Authorization boundary: The point at which a recommendation becomes an action requiring the buyer’s permission.