Proof-Based Pricing: Why Unverified Claims Are Worth Nothing to an Algorithmic Buyer

Source abstract

Pricing theory has, until now, been written for a buyer who can be persuaded. Its three orientations, cost-based, competition-based, and customer value-based, differ in what they anchor a price to but agree on who reads it: a human who can be informed, framed, and convinced. In a growing, new class of consumer markets the reader is an AI agent that admits, verifies, and computes under an evaluation rule set by an intermediary, and the seller's claims reach that rule, not a person. This note defines Proof-Based Pricing: a seller-side pricing model for this configuration, in which the seller decides which claims to substantiate, how to make an offer eligible, and how to price when a delegated buyer applies an externally specified evaluation rule. The premise, that an algorithmic buyer discounts what it cannot verify and rewards what it can, was first set out in the author's book The Algorithmic Shopper (manuscript registered with the US Copyright Office in July 2025; St. Martin's Press, 2027); the pricing model itself is defined here. The closest formal ancestor is the procurement scoring auction, in which a buyer selects among bids under a published rule; Proof-Based Pricing adapts that selection to consumer markets in three structural ways: the evaluation rule is set by an intermediary exogenous to both sides; admission turns on the machine legibility of the offer; and the seller responds with a published price schedule on verifiable attributes rather than a bid. We characterize acceptance by the proof coefficient, the agent's posterior expected value relative to the schedule price: an admitted offer is acceptable exactly when the coefficient is at least the intermediary's price weight, with equality for the marginal offer. We state five falsifiable propositions with the observation that would refute each, set out a measurement agenda with named test beds, and specify the first experiment.