Outcome-based pricing for AI software: advantages, limits and nuances according to a Constellation Research panel

The article, published by Constellation Research and based on a panel at its Connected Enterprise 2026 conference, looks at outcome-based pricing. Its central argument is that the idea sounds appealing but is hard to put into a contract. Even so, enterprise buyers have a real interest in sharing risk with vendors. The author argues that the most likely end point is a hybrid model, with a base fee, a variable component tied to outcomes and milestone payments.

The panelists were Anil Cheriyan (Phase IV Ventures), Minerva Tantoco (City Strategies LLC) and Alisha Thompson (Team Air Distributing). The panel came after a workshop on how to align organizations around outcomes.

First takeaway: outcomes are not the same as metrics. Tantoco explained that outcomes are business results, such as value created, productivity or customer satisfaction. Thompson linked them to removing friction and pain points for customers and teams, and to making life and work easier. Tantoco added that defining outcomes too narrowly at the start can hide unexpected benefits.

Second takeaway: putting a price on outcomes is hard. Cheriyan noted that outcomes are difficult to measure and to attribute to a single company. Even cost savings, he said, are disputed within the buyer's own organization. Thompson said vendors generally avoid guaranteeing outcomes. They would rather take responsibility for successful software installations and formal sign-off. Tantoco warned that any single metric can be gamed. One example is case resolution time, which could "improve" with a system that simply turned everyone away. Processes with clear metrics and an established baseline are the best fit for these kinds of deals.

Third takeaway: buyers are open but cautious. Thompson said she wants to see shared risk and reward tied to an outcome. Tantoco advised companies to model the best-case scenario first, so they know how much they would have to pay if the project is very successful.

Fourth takeaway: services firms may be the best fit. Cheriyan described a recent business process outsourcing deal in which the provider stopped charging per person and charged per transaction instead. For these firms, outcome-based pricing could mean accepting lower revenue in exchange for a share of AI-driven efficiency gains and, possibly, some additional upside.

Obstacles remain. Cheriyan said attribution is still difficult and procurement teams need training, because their default stance is to reject non-standard contract structures. For now, the answer is "it depends." Thompson summed it up by saying everything comes down to the specific deal: how much the vendor has at stake, what counts as the success metric and how much risk each party takes on. The deal also has to be put in writing, and both parties must share the risk. Just because something hasn't been seen before, she said, doesn't mean it can't be beneficial.

The piece also links to other coverage from the firm on outcome-based pricing at vendors such as Oracle, HubSpot, Salesforce and Pegasystems. However, the downloaded page shows only the headlines and not the articles themselves, so no conclusions can be drawn from them.