When an AI shopping assistant recommends a product based on a Wirecutter review, who gets credit for the sale? Today, the answer is: nobody knows. The content creator has no visibility, the merchant can't attribute the conversion, and the AI agent operates as a black box.
Issue #185 proposes org.openattribution.telemetry — a UCP vendor extension that embeds content citation data directly into checkout sessions. This isn't about tracking users; it's about tracking which content influenced their decisions.
Alex Springer's proposal articulates a market failure that threatens the AI commerce ecosystem:
Content creators — reviewers, guide authors, comparison sites — invest significantly in producing the expert content that AI shopping agents rely on to generate recommendations. This content is the raw material of AI-assisted commerce: without reviews, there are no credible recommendations; without guides, there is no informed comparison.
The issue is stark: if creators can't measure whether their content drives purchases, they have no business case for allowing AI agents to access it. The rational response is to restrict access — which threatens the content supply chain that makes AI commerce valuable in the first place.
The extension adds an attribution object to UCP checkout sessions, recording which content was retrieved and cited during the conversation:
{
"id": "chk_123",
"line_items": ["..."],
"attribution": {
"content_scope": "electronics-reviews",
"prior_session_ids": ["550e8400-e29b-41d4-a716-446655440999"],
"content_retrieved": [
{
"content_url": "https://www.wirecutter.com/reviews/best-wireless-headphones",
"timestamp": "2026-01-15T10:30:01Z"
}
],
"content_cited": [
{
"content_url": "https://www.wirecutter.com/reviews/best-wireless-headphones",
"timestamp": "2026-01-15T10:30:05Z",
"citation_type": "paraphrase",
"excerpt_tokens": 85,
"position": "primary"
}
],
"conversation_summary": {
"turn_count": 3,
"primary_intent": "comparison",
"topics": ["headphones", "noise-cancelling"],
"total_content_retrieved": 5,
"total_content_cited": 2
}
}
}The extension defines four granularity levels for data sharing:
This addresses the privacy tension — merchants and content creators can negotiate appropriate data sharing levels without compromising user privacy.
Attribution telemetry provides something that doesn't exist today: measurable ROI for AI-accessible content. Wirecutter could know that their headphone review influenced 10,000 purchases last month — data that justifies content investment and enables partnership negotiations.
Merchants gain insight into which content partnerships drive conversions. If TechRadar reviews convert better than CNET reviews, that informs content strategy and affiliate relationships.
Without attribution, the content ecosystem risks collapse. If creators can't demonstrate value, they'll block AI access. If merchants can't measure content ROI, they won't pay for partnerships. Attribution telemetry is the feedback loop that makes the system sustainable.
contradiction as a citation type — content that was retrieved but explicitly disagreed with. This captures cases where AI agents use content to justify not recommending a product, which is also valuable signal for creators and merchants.
The proposal addresses a critical concern: what stops an AI agent from claiming false citations to inflate attribution?
Several mitigations are built in:
The key insight: the extension provides attribution signals, not attribution decisions. Consumers apply their own confidence weighting and fraud detection.
This proposal builds on Issue #180: Attribution Signals Extension from two days ago. Where that proposal focused on merchant-side conversion attribution, OpenAttribution Telemetry tackles content-side influence attribution. They're complementary — a complete attribution picture requires both.
The proposal is more than a concept document. According to the enhancement proposal:
This level of completeness suggests serious intent. The OpenAttribution organization appears purpose-built for this problem.