Methodology

How GradeMY evaluates AI shopping readiness.

GradeMY reads public storefront information available to an AI shopping system, tests whether important buying facts can be verified, and turns observed blockers into specific remediation.

Public storefront scan · Read-only · No checkout, cart, account, payment, customer, order, or admin access.

1. Collect the public storefront surface

The scan fetches accessible public pages and metadata. Depending on what a store exposes, evidence can include storefront HTML, product and collection paths, product cards, product detail pages, Product and Offer structured data, prices, currency, availability, variants, identifiers, images, robots.txt, sitemaps, and public merchant trust or policy links.

The scan is bounded and sampled. It is not a whole-catalog certification.

2. Test a shopping request

A visitor can optionally provide a specific shopping request. GradeMY compares its constraints with captured evidence: whether the product can be discovered, identified, priced, matched to a variant, and confirmed available. The report records task completion and the point where evidence becomes insufficient.

If no custom request is supplied, the report still evaluates public product and offer readiness from the sampled pages.

3. Normalize evidence and blockers

Customer-facing remediation comes from the canonical report. Findings retain evidence provenance and defect granularity. A missing Offer.availability field remains an availability defect; it is not widened into a generic Product + Offer rebuild unless the evidence establishes that the broader schema is absent.

Capabilities that were not tested are labeled not evaluated. GradeMY does not manufacture a failure—or a pass—from missing evidence.

4. Produce a developer-ready handoff

When an actionable finding is supported, GradeMY can provide a recommended fix and developer ticket with an acceptance test. Completed scans expose only the artifacts they actually emitted: browser report, HTML report, PDF, and developer tickets when tickets exist.

A fresh rerun repeats collection and evaluation against new public evidence. It is proof that the observed surface changed, not a guarantee of rankings, recommendations, traffic, or revenue.

5. Put agent interfaces in context

Agent Commerce

GradeMY uses Agent Commerce to describe product discovery and evaluation by AI shopping systems. The current public scan stops before cart, checkout, payment, account, and order placement.

AEO and GEO

Answer Engine Optimization and Generative Engine Optimization are adjacent contexts. GradeMY does not claim general AEO/GEO ranking performance; it evaluates the observable product evidence needed for a shopping decision.

WebMCP

WebMCP is reported only where its surface was evaluated. An experimental, absent, or untested interface is not treated as evidence of transaction capability and does not create an unrelated product-evidence repair.

6. Interpret the result within its limits

  • Results describe sampled public evidence at scan time.
  • Storefront rendering, access controls, or network failures can limit collection.
  • GradeMY does not access private store or customer systems in a default scan.
  • GradeMY does not guarantee inclusion, placement, recommendation, ranking, traffic, or revenue in any AI or search system.
  • Unsupported transaction protocols or capabilities are not marketed as live.