# Outcome-Based Pricing
Pricing that ties the bill to a measurable business outcome (revenue recovered, ticket resolved, cost avoided) rather than access (seats) or volume (usage). The structural response to AI compressing the cost of work. See [[AI Eats Services Not Software]].
## Three emerging patterns
1. **Enterprise bundle** — recurring platform license plus a fee for a human deployment / success team. Familiar to enterprise buyers; preserves seat mechanics underneath.
2. **Outcome tier** — standard tier for raw AI output, premium tier for guaranteed resolutions with human backing. Aligns price to certainty of result.
3. **Fractional employee** — priced as a share of a full-time salary the AI replaces. Implicit admission that the thing being sold is *work*, not *software*.
## Requirements for outcome pricing to work
- The outcome is **measurable** cleanly and in a way the buyer trusts.
- The measurement window is **short enough** that disputes resolve before churn.
- The **delivery cost** (inference, human loop, infra) is well-modelled; otherwise the company absorbs tail risk on the cost side while passing upside to the customer.
- The buyer is **economic**, not technical — only line-of-business P&L owners buy outcomes.
## Failure modes
- Outcomes that can't be attributed to the product get contested at renewal.
- Inference cost spikes erase gross margin on the biggest customers.
- Customers shape behaviour to game the metric.
Related: [[AI Eats Services Not Software]], [[Industrial AI Unit Economics]], [[AI Agent Vertical SaaS DD MOC]], [[Investing/Investment Thesis Structure]]
---
Tags: #businessmodel #AIstrategy #investing