The framework behind the evidence layer
How Cartograph is specified to observe merchant surfaces, how evidence is labeled, and exactly where the scope stops.
Readable, Consistent, Transactable.
This is a readiness framework, not an ACES taxonomy, classifier, confidence model, grade, or score. Each category names the merchant surface, the observation method, and its limitation.
Can the defined scanner or collector fetch and parse the named merchant surface?
Boundary: not external-model understanding, citation, preference, or recommendation.
Do named merchant-authorized surfaces materially agree within a defined observation window and comparison policy?
Boundary: not external-agent trust, belief, confidence, preference, rejection, or consideration-set placement.
Do merchant-observed events or a separately authorized synthetic diagnostic show progression through the exact cart or checkout milestones in scope?
Boundary: not complete-journey proof, order completion, payment authorization, provider support, or conversion.
Evidence, not guesswork.
The v0.1 specification defines deterministic traffic labels and evidence codes. It does not infer intent, score actors, or treat missing agent evidence as proof of human traffic.
The v0.1 specification defines five traffic labels: verified agent, declared bot — verified, declared bot — unverified, AI-referred, and unclassified. Those labels describe available evidence at capture; they do not assert that unclassified traffic is human.
Know what the evidence supports
- Verified
- Identity supported by the applicable verification evidence.
- Claimed
- A request declares an identity that has not been verified.
- Unknown
- The available evidence does not establish an identity.
Unknown does not mean human. An AI-referred visit is a referral signal, not proof of autonomous shopping.
Technical labels and definitions (v0.1 specification)
The v0.1 specification uses five traffic labels:
- verified_agent
- declared_bot_verified
- declared_bot_unverified
- ai_referred
- unclassified
Unidentified Traffic describes eligible storefront requests that remain unclassified after deterministic labeling. It is classification uncertainty, not a measured shortfall in another analytics tool.
Unidentified Traffic is the percentage of eligible storefront request events that remain unclassified after deterministic labeling. It is measurable only with complete server/CDN/origin log coverage.
- Numerator
- Eligible log-origin request events labeled
unclassified. - Denominator
- All eligible log-origin request events across the five v0.1 labels.
- Coverage requirement
- Complete server/CDN/origin log coverage.
- Coverage failure
- Not measurable — incomplete log coverage.
- Interpretation
- Not human share, agent share, readiness, conversion impact, or economic loss.
- Future receipt
- Each result carries a measurement window, a denominator count, a coverage status, a label version, and an exclusion-policy version.
From an observation to a useful next step.
Product preview: this describes the intended pilot scope, not shipped integrations.
- 01
Agree the scope
Supported data sources, setup, permissions and terms are agreed before anything is connected.
- 02
Review the evidence
Requests, identity evidence and shopping milestones, where the data supports them. Gaps stay visible instead of being filled in.
- 03
Choose what to investigate
Review specific responses, pages or data with your developer. Any change to your store stays your team’s decision.
Register interest — interest only; it does not grant product access or start a pilot.
Honest about the scope.
- ×We do not replace your ecommerce platform.
- ×We do not replace your PSP.
- ×We do not replace your PIM.
- ×We do not blindly allow agents to bypass checkout, fraud, or payment controls.
- ×We do not claim every AI crawler is trustworthy.
- ×We do not ask for sensitive customer or payment data for the initial scan.
Questions worth having answered.
- ›Can the defined scanner or collector fetch and parse our named product surfaces?
- ›Do our product pages, feeds, structured data, and policies materially agree within the defined observation window?
- ›What was the last merchant-observed cart or checkout milestone?
- ›Which challenge or block response, if any, was recorded at a named milestone?
- ›Which eligible request events remain
unclassifiedafter deterministic labeling? - ›Which request-evidence fields are absent from our current analytics view?
The infrastructure for agentic commerce is forming now.
Product feeds, checkout protocols, delegated payments, agent identity, crawler controls, and referral analytics are relevant merchant-side surfaces as agent commerce develops. Cartograph is being built to document evidence from those surfaces without claiming what an external model understood, preferred, or decided.