The inputs to impact chain is a five-step model from program evaluation that Potio uses to name pricing models by where they bill: inputs, activities, outputs, outcomes and impact. Each step up the chain moves the invoice closer to the customer’s real result, and each step is harder to measure than the one before.
| Step | What it is | Example | Pricing model |
|---|---|---|---|
| Inputs | Resources consumed | Tokens, API calls, gigabytes | Usage-based pricing |
| Activities | Work attempted | Agent runs, tool-call sequences | Agentic pricing |
| Outputs | Work completed | Resolved ticket, qualified lead | Output-based pricing |
| Outcomes | Change in the customer’s business | Cash recovered, fraud loss reimbursed | Outcome-based pricing |
| Impact | Change in the world | Lives saved, market moved | Nobody bills here |
A support AI shows the difference. Billed on inputs, the customer pays for twelve turns of inference even if the chat ends in an escalation. Billed on outputs, the vendor eats those twelve turns and invoices only resolved conversations. Billed on outcomes, the vendor would need proof the resolution kept a customer or saved money.
Potio’s take: Nearly every vendor marketing outcome-based pricing is billing on outputs. That is often the right call. The rule is to move as far toward the outcome as your value metric survives, then stop, and never trade metric quality for proximity to the outcome.
Potio borrowed the chain because software pricing uses one word, “outcome”, for two different things. Intercom’s Fin at $0.99 per resolved conversation, Zendesk’s verified resolutions and Sierra’s per-session pricing are all outputs. Each is a state change inside the vendor’s own software. None measures money moving in the buyer’s business.
Real outcome pricing survives only where three things hold: someone outside both companies decides whether the result happened, cash actually lands in the customer’s account, and the vendor eats the full cost of failure. Chargeback recovery, where card networks adjudicate each dispute, passes. Very little else does.
Impact is where founders go when they skip steps. The parking enforcement example, discussed in a public pricing forum, billed per violation processed and then asked about charging for lives saved. That is a jump from outputs straight to impact. Everything workable sat in between.
In practice most AI companies land on a hybrid: a platform or seat floor, plus a variable layer at the furthest step their metric can defend.
At activities. The customer pays for the agent’s work attempted, whether or not it produced a completed output, so the buyer carries the risk of failed runs.
No. Impact depends on too many factors outside any one vendor’s control to be measured or attributed at invoice time.
No. A resolved ticket is an output, a completed piece of work the software can log. An outcome would be the business result that resolution produced.
More on this: Outcome-Based Pricing: Why It Rarely Works in SaaS
I'm a 3x founder and former CEO of Toggl. I work hands-on with SaaS & AI teams to fix pricing, packaging and monetization.
Book a call