Methodology

How the agentic readiness score is calculated.

The full method: what is checked, how each category is weighted, how the overall score is produced, and — the part most vendors leave out — what an automated scan cannot see.

Two instruments

One framework, two depths.

The free audit reads your public storefront exactly the way an agent does. The expert assessment reads everything an agent would reach if you let it in.

Free · Automated

Agentic Readiness Audit

Scope: Public storefront pages
Checks: 199 automated checks, 10 weighted categories
Turnaround: Under a minute
Fee: None. No account required
Output: Score out of 100, category breakdown, prioritized fixes

Run the Free Audit

Free · Expert review

Readiness Assessment

Scope: Platform, APIs, feeds, checkout, analytics
Checks: 126-point manual review, all 7 dimensions
Turnaround: 3 business days
Fee: None, and no obligation to engage further
Output: Scored report with a costed remediation roadmap

What it covers

Scoring

Four statuses, two formulas.

Every check returns one of four statuses. Not-applicable checks are excluded from the average entirely — a site is never penalised for a feature it was never expected to have.

Pass: The signal is present and valid.
Warn: Present but incomplete, ambiguous, or below best practice.
Fail: Absent or broken from an agent's point of view.
Not applicable: Nothing to assess. Excluded from the score.

The formulas

category score = mean(applicable check scores) × 100

overall score = ∑ (category score × category weight)

Weights are fixed and sum to 1.0, so the overall score is directly comparable between scans, between competitors, and over time. A category you cannot fix does not quietly change the weighting of the ones you can.

Category weights

What counts, and how much.

Weighting follows commercial impact. Checks that decide whether an agent can complete a transaction outrank checks that only affect how likely you are to be cited.

AI Agent Tools & Action Surfaces

0.18

Machine-readable actions an agent can call: agent manifests, AI catalog metadata, declarative forms, tool endpoints, cart and checkout surfaces.

Pass: a published agent catalog describes your product and checkout actions. Fail: the only path to a purchase is a JavaScript click handler.

Content Discoverability

0.15

Whether an agent can find and fetch your content at all: sitemaps, canonical URLs, render mode, pagination, feed exposure, llms.txt.

Pass: product content is present in the server response. Fail: content only appears after client-side hydration.

Structured Data & Schema Markup

0.12

JSON-LD coverage and validity for Product, Offer, Organization, FAQ, Breadcrumb, Review, and their required fields.

Pass: Product with price, priceCurrency, and availability. Fail: Product markup with no Offer node.

Technical Readiness & Security

0.09

Response codes, redirects, TLS, headers, rate limiting, and whether agent traffic is blocked by bot mitigation.

Pass: agent user-agents receive 200s. Fail: a challenge page is returned to non-browser clients.

AI Crawler Permissions

0.08

robots.txt and header directives for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and other AI crawlers.

Pass: named AI crawlers are explicitly allowed. Fail: a blanket disallow that silently removes you from the answer layer.

Meta Tags & AI Head Elements

0.08

Titles, descriptions, Open Graph, product meta, language and locale signals, and head-level hints models read.

Pass: unique, specific titles per product. Fail: one templated title across the catalogue.

Semantic HTML & Content Structure

0.08

Heading hierarchy, landmarks, lists, tables, and whether key commerce facts sit in parseable elements.

Pass: price and availability in text nodes. Fail: price rendered as an image or background sprite.

Generative Engine Optimization

0.08

Whether your content is quotable: entity clarity, claim specificity, comparison coverage, citation-ready formatting.

Pass: a definition a model can lift verbatim. Fail: a page of adjectives with no extractable claim.

Accessibility & Agent Interaction

0.07

ARIA roles, labels, focus order, and form semantics — the same signals an agent uses to operate a page.

Pass: labelled inputs on the add-to-cart form. Fail: unlabelled inputs an agent cannot map to fields.

Answer Engine Optimization

0.07

Question-shaped content, direct answers, FAQ structure, and the formats answer engines cite.

Pass: a question heading followed by a one-paragraph answer. Fail: the answer buried in paragraph nine.

Readiness tiers

What the number means.

Tier thresholds are fixed, so the same site scanned twice is directly comparable. A low first score usually means agent action surfaces and crawler permissions were never configured — not that the site is poorly built.

90–100
Agent Ready: Agents can find, read, and transact. Remaining work is optimization.
70–89
Partially Ready: Agents can read you, but at least one path to purchase breaks.
50–69
Needs Work: You are visible in places and invisible in others. Gaps are structural.
0–49
Not Ready: Agents cannot reliably read or act on your storefront.

Coverage

What the automated scan cannot see.

The audit covers the seven dimensions of agentic readiness unevenly, because a public scan only reaches public surfaces. Here is exactly where it stops.

AI Discoverability: Fully covered; Content Discoverability, Crawler Permissions, Structured Data, Meta Tags, AEO, GEO.
Protocol Readiness: Partly covered; The scan detects public manifests and agent action surfaces. It cannot see private UCP, ACP, MCP, or A2A endpoints, or negotiate a session.
Data Quality: Partly covered; The scan reads what your pages publish. Feed completeness, attribute accuracy across the full catalogue, and inventory sync require access to the feed and platform.
Transaction Enablement: Partly covered; Checkout surfaces are detected; payment handler exposure and session security are reviewed manually — the scanner never completes a purchase.
Trust & Verification: Partly covered; Review and policy markup is checked. Merchant verification status and returns handling are reviewed manually.
Post-Purchase Experience: Not covered; Order status APIs, tracking, and returns sit behind authentication. Expert assessment only.
Measurement & Optimization: Not covered; Agent session attribution and channel analytics require access to your analytics stack. Expert assessment only.

Full definitions of each dimension are in the 7-dimension framework.

Questions

Method questions.

How is the agentic readiness score calculated?

Every check returns pass, warn, fail, or not-applicable. Not-applicable checks are excluded so a site is never penalised for a feature it was never expected to have. Each category score is the mean of its applicable check scores, expressed 0–100. The overall score is the sum of each category score multiplied by its fixed weight — Agent Tools 0.18, Content Discoverability 0.15, Structured Data 0.12, Technical Readiness 0.09, Crawler Permissions 0.08, Meta Tags 0.08, Semantic HTML 0.08, Generative Engine Optimization 0.08, Accessibility 0.07, Answer Engine Optimization 0.07. Weights sum to 1.0.

Why is Agent Tools weighted highest?

Because it is the only category that decides whether an agent can act rather than merely read. A brand can be perfectly described and still lose the order if there is no machine-callable path to a cart. Discoverability determines whether you are named; agent tools determine whether you are bought.

What score should a store aim for?

90 or above is Agent Ready. 70–89 is Partially Ready — models can read you but at least one purchase path breaks. 50–69 is Needs Work. Below 50 is Not Ready.

What can the free audit not see?

It reads your public storefront the way an agent does, so it cannot evaluate private protocol endpoints, full-catalogue feed quality, payment handler configuration, post-purchase APIs, or your analytics attribution. Those are the areas the 126-point expert assessment adds.

How does the audit differ from the expert readiness assessment?

The audit is an automated scan of public pages, free, scored in under a minute. The expert readiness assessment is a 126-point manual review of your platform, APIs, data feeds, and configuration over three business days, delivered as a scored report with a prioritized, costed remediation roadmap.

Run the Free Audit See a Real Report

Same method, your domain, under a minute.