Outcome Pricing

Outcome-based pricing moves the value metric downstream toward the result the customer is buying, and moves execution risk along with it.

A vendor billing per seat is paid for access. A vendor billing per resolved conversation is paid for finished work, and has agreed that unfinished work is free. That trade lives in the licensing model, where the value metric is chosen. The rate on the pricebook is downstream of it.

Metrics sit on a line that runs from the vendor's own activity to the customer's realized result. Each step down that line gets closer to something the buyer already recognizes as value, and each step takes on more of the customer's execution risk.

Charge on tickets created and the buyer may argue the count has nothing to do with what they got. Charge on tickets resolved for six months and the vendor is accountable for whether the customer's team followed through. The designs that hold up land between those poles, and finding that point is the work.

The argument that arrives in month four is about whether the outcome happened and who gets to say so. The rate on the contract is rarely what is in dispute.

Every outcome metric is a proxy for something the vendor cannot observe directly, and the distance between the proxy and the real result is where the disputes live. A resolution counter sees a session that closed. It does not see the customer who came back the next morning with the same question, or the case the software handed quietly to a human.

Four things have to be settled before a result can carry a price. What counts as the result. Who measures it. What happens when a measurement is challenged. What the vendor is paid for work that lands outside the count.

One question sits upstream of all four. When the result is commingled, produced by the vendor's software alongside the customer's own people and two other systems, the buyer has a standing invitation to argue the vendor's share down at every renewal. Attribution is a licensing question before it is a billing question, and it is answered in the architecture or it is answered in the negotiation.

Start with the overview below: the definition, the performance-contracting history it inherits, and the test that decides whether a result can carry a price. Each read after it takes one question. How the model gets designed once the metric is chosen. What a per-resolution meter counts in practice. How the vendors already selling this way have written their terms.

The dated record behind those reads lives in the SPP Pricing Observatory, where vendor pricing moves are verified, dated, and maintained as arcs. Articles here render that ledger rather than restating it, so the evidence stays current as the ledger is re-verified.

Sibling hubs carry the neighbouring questions. Credits, consumption, and the wider AI economics core live in AI Pricing, including the hybrid shape that pairs a base fee with an outcome fee and the credit wrapper that hides the metric. Who pays for an agent run that produces nothing useful is worked through in Harness Pricing, at When the Agent Is Wrong.

If that decision is live for a product already in market, describe it through AI pricing strategy and a pricing expert who has architected outcome contracts will reply.