Nobody Knows What "Outcome-Based Pricing" Actually Means
Outcome-based pricing is the pricing idea that sounds perfect in a pitch deck and becomes a legal nightmare the moment you try to define the contract terms. Every AI company announcing it as their go-to-market strategy is telling you something important: they haven't tried to collect yet.
The pitch is elegant. Instead of charging for access, or usage, or seats — you charge for results. The AI closes a deal? You take a cut. The agent deflects a support ticket? You get $0.99. The automation saves 30 hours of analyst time? You bill for the hours saved. It's the ultimate alignment: you win when the customer wins. No win, no payment.
This is theoretically beautiful. In practice, it runs into a problem so old it predates software: who decides what counts as a result?
The Attribution Problem Is Not a Small Problem
Intercom charges $0.99 per "resolved" support ticket. Sounds clean. But "resolved" is doing a lot of work in that sentence. Resolved by whom? If a customer submits a ticket, the AI responds, the customer doesn't reply for 48 hours, and the ticket auto-closes — is that a resolution? What if the customer reopened it two days later? What if the AI gave an answer and the customer figured it out themselves, ignoring the AI entirely? Intercom gets to decide. And your legal team is about to read that contract very carefully.
BCG mapped five agentic AI pricing models in 2025: resource-based, agent-based, interaction-based, outcome-based, and hybrid. Four of the five have some form of attribution ambiguity. The only clean one is resource-based (pay for compute consumed), which is just usage-based with extra steps. Every model that ties payment to business outcomes requires you to answer: how do we know the AI caused this?
Consider a sales AI that "generates qualified leads." A rep closes a deal six weeks later. How much credit does the AI get? It found the lead. But the rep worked the account, customized the pitch, and handled the objections. Is this a 100% AI win? 50%? 20%? This is not a pricing question. It's a philosophy of causation question. And you're trying to close enterprise deals with it.
Why It Mostly Becomes Hybrid Anyway
What actually ships in the market — as opposed to what gets announced at conferences — is almost always a hybrid model with an outcome component bolted on. A platform fee establishes a floor. Usage metering tracks activity. And there's a performance kicker tied to outcomes: you pay the base either way, and if the AI hits defined targets, there's a bonus component.
This structure survives procurement review for three reasons:
- The vendor absorbs less risk (they get paid something regardless)
- The customer accepts it because the platform fee is visible and predictable
- The outcome component is structured as a bonus, not the core model — easier to audit, easier to dispute, easier to remove in year two
Pure outcome-based pricing — where zero base fee means zero guaranteed revenue — works in verticals with extremely clear, auditable, binary outcomes. Fraud detection: the fraud was either caught or it wasn't. Document processing: the document was either extracted correctly or it wasn't. Route optimization: the routes either cost less than baseline or they didn't. These are well-defined.
If you can't write a two-sentence definition of "success" that both your CEO and your customer's CFO would sign off on, outcome-based pricing will eat your renewal.
What You Actually Need Before Attempting This
Before announcing outcome-based pricing, make sure you have: (1) a measurable baseline agreed upon in writing before deployment, (2) attribution rules that both parties sign before go-live, not after, (3) a data pipeline that generates an auditable outcome report the customer can independently verify, and (4) a dispute resolution mechanism that doesn't require litigation.
Most AI companies announcing outcome-based pricing have zero of these. What they have is a pricing page that says "pay only for results" and a sales deck that defines "result" in a font size of approximately 8pt.
Outcome-based pricing is the future. The present is mostly vibes and extremely flexible outcome definitions. When the audits start coming in, the contracts are going to get very creative.
Sources
- Orb: Pricing AI Agents — 4 B2B Pricing Models for 2025 — outcome-based design principles and attribution rules
- BCG: Agentic AI Pricing Models (2025) — five model taxonomy; resource-based, outcome-based, hybrid analysis
- Intercom Pricing — $0.99/resolved ticket outcome model in production
- Aakash Gupta: Ultimate Guide to B2B SaaS Pricing (2024) — outcome-based vs. value-based distinctions