PMaker home
Once pricing is tied to outcomes, delivery must define what success meansBind to outcomePay for the result receivedBoth sides alignUnused features lose priorityMeasure earlyLock acceptance metrics before workValue is testableBuild the exam firstDeepen the accountExpand from one team to a networkValue creates a handleUse data to open doorsOutcome pricing is powerful and honest: delivery must answer whether the customer succeeded

Customers pay for what they receive, so delivery must define success before it starts.

Charge for Outcomes

When payment is tied to outcomes, the delivery team’s priorities and behavior change with it.

a16z lists charging for outcomes rather than licenses — “outcomes, not licenses” — as one of the four features of “Palantirization.” Once price is tied to results, the delivery team’s behavior is reordered.

After reading, you should be able to answer:

  • Why charging for outcomes is the most powerful commercial model and the most dangerous one
  • How to define a “result” as a metric both sides accept and can measure
  • Why outcome-based pricing forces you to actually stand on the customer’s side of the table

Bind the work to the result

If the team is paid for a resolved conversation, a completed workflow, or a measurable reduction in waste, nobody wants to spend three weeks polishing a feature nobody uses — an unused feature is now a cost the delivery team carries. Palantir’s early government work already leaned on pay-only-if-it-gets-done terms; Sierra charges per resolved conversation, and an unresolved one bills nothing. a16z writes the sequence in one line: first you earn the seat and the trust at the top of the customer’s organization, you deliver the result, and only then do you price on outcomes a16z-palantirization. Outcome pricing brings the customer and delivery team onto the same side of the table.

Put measurement first

The first step of an FDE project is not code. It is agreeing on what “done” means. A bootcamp-length window (see Minimum Viable Deployment) forces one sharply focused target, such as “cut scheduling conflicts by 30%”, not “explore AI for manufacturing”.

Morgan Stanley took the order seriously when it adopted large models: build the evaluation first, then ship the model. Historical question-and-answer pairs became a scoring set, retrieval hit-rate became the gate, nothing was rolled out until it cleared, and only then did the assistant appear on advisor desks step by step openai-morgan-stanley. Writing the exam before iterating the answer matters because it turns “what counts as done” into something neither side can renegotiate afterwards — which is the precondition for charging by outcome at all.

Deepen the existing account

Once one customer proves value, expand from one department to a network. Usage data opens the next conversation with the group or the next site. Every dollar tied to a visible outcome gives the expansion a concrete handle.

References

  1. The Palantirization of everything
  2. Morgan Stanley uses AI evals to shape the future of financial services