How the score works

The Agent-Readiness Score is a weighted roll-up of four layers, each scored 0–100 from individual checks. Every check contributes evidence we capture at scan time — we never report a finding we can't show proof for.

Discoverability

35%

Whether AI systems can find the store and enumerate its catalog: crawler access, sitemaps, feeds, and machine-readable product data. Weighted highest because nothing else matters if agents can't find you.

Comprehension

25%

Whether machine readings of the store agree with each other and with reality: price, stock, and policy consistency across every source an agent might read.

Transactability

25%

Whether a purchase journey is mechanically completable: search, product configuration, cart, and checkout reachability for automated buyers.

Protocol readiness

15%

Whether the store speaks the emerging agent-commerce protocols. Weighted lowest today, and re-weighted as adoption grows — the rubric is versioned.

Philosophy: weights reflect where agent purchases actually fail today, and shift as the ecosystem matures — each report carries its rubric version. Within layers, checks are weighted by how often the underlying pattern breaks real agent journeys. We publish categories and weighting philosophy, not check implementations — partly to keep the score hard to game, and partly because the implementations change as agents do.