The AI Commerce Readiness Standard
ACRS v1.1 — how a website is scored for whether AI systems can find it, understand it, and transact with it. Three tiers, 22 automated checks, one 100-point score.
How scoring works
- 100 points across 22 automated checks → one letter grade.
- Tier badges are earned by passing every check in a tier — the progressive ladder (Discoverable → Readable → Agentic Commerce Ready).
- Tier 3 is two-track: honest declarative markup earns Agentic-Ready (declared); a real, callable agent endpoint earns Agentic-Verified.
- Each check scores pass / partial / fail = full / half / none of its points.
| Grade | Score | Meaning |
|---|---|---|
| A | 90–100 | Fully agentic-commerce ready |
| B | 75–89 | Strong readiness, minor gaps |
| C | 55–74 | Moderate readiness, key gaps |
| D | 35–54 | Basic presence, significant work |
| F | 0–34 | Not AI-commerce ready |
Tier 1 — AI Discoverable (35 pts)
Can AI crawlers reach, fetch and identify the site?
| ID | Check | Pts | What it verifies |
|---|---|---|---|
| T1.1 | HTTPS / TLS | 5 | Valid cert, HTTP→HTTPS redirect, HSTS header. |
| T1.2 | Identity schema | 6 | LocalBusiness / Organization JSON-LD with name, url, logo, sameAs. |
| T1.3 | Offer / Service schema | 6 | Product+Offer or Service with an offer/price. |
| T1.4 | AI answer-bot access | 6 | robots.txt doesn't block the answer/agent bots. |
| T1.5 | sitemap.xml | 4 | Valid sitemap, referenced in robots.txt. |
| T1.6 | Core Web Vitals | 5 | LCP ≤ 2.5s, CLS ≤ 0.1, INP ≤ 200ms (mobile). |
| T1.7 | Mobile responsive | 3 | Viewport meta; usable at 375px. |
Tier 2 — AI Readable (40 pts)
Can an AI accurately describe, price and recommend the business?
| ID | Check | Pts | What it verifies |
|---|---|---|---|
| T2.1 | llms.txt | 6 | Present, 200+ words, with description and contact. |
| T2.2 | AI-usage policy | 4 | Declared via robots.txt, Content-Signal, or ai.txt. |
| T2.3 | FAQ schema | 5 | FAQPage, 5+ Q&As, 30+ words each. |
| T2.4 | Pricing signals | 6 | Machine-readable price / priceRange / PriceSpecification. |
| T2.5 | Service / product pages | 6 | One page per offering: single H1, 300+ words. |
| T2.6 | Availability & area | 5 | openingHours, areaServed, availability. |
| T2.7 | Trust signals | 5 | AggregateRating/Review or a verifiable accreditation. |
| T2.8 | Accessibility baseline | 3 | Automated WCAG 2.2 AA basics pass. |
Tier 3 — Agentic Commerce Ready (25 pts · two-track)
Can an AI agent actually act — enquire, book, or transact? Declarative markup earns Agentic-Ready (declared); a real callable endpoint earns Agentic-Verified.
| ID | Track | Check | Pts | What it verifies |
|---|---|---|---|---|
| T3.1 | Declarative | Action schema | 4 | ContactAction/ReserveAction/OrderAction with a real target URL. |
| T3.2 | Declarative | Service catalogue | 4 | ItemList or 3+ Service/Product entities with offers. |
| T3.3 | Declarative | Contact / booking | 4 | contactPoint + a potentialAction an agent can follow. |
| T3.4 | Functional | Agent-surface manifest | 6 | A working rail — UCP, ACP feed, MCP, A2A, or an OpenAPI action endpoint. |
| T3.5 | Functional | Feed ↔ schema consistency | 3 | Product feed matches on-page Offer schema. |
| T3.6 | Functional | Published SLA | 2 | A machine-readable response/fulfilment commitment. |
| T3.7 | Functional | Verified identity | 2 | Registration/VAT number + resolving sameAs. |
Open & verifiable
ACRS is published openly — there's no black box. The full rubric, with every check's exact pass condition and weighting, is machine-readable at /api/v1/acrs/rubric. Sites are scored by the futureproofwebsite.com audit engine; partners can request API access to the scan endpoint. See the API for how agents and integrations call it.
Who maintains it
ACRS is developed and maintained by Datavision, and the scoring engine runs on futureproofwebsite.com. The standard is intended to be implementation-neutral: the rubric is open, so any tool can score against it and any site can self-assess.