B2B AI Pricing Shifts to Outcome-Based Models

The shift away from per-seat pricing is driven by a simple reality: one AI agent can potentially do the work of ten or even a hundred users. As agentic AI makes software more autonomous, tying price to human users becomes obsolete, forcing vendors to align cost with the work being done. Intercom's "Fin" AI agent exemplifies this with a pure pay-per-resolution model at $0.99 per solved ticket. A resolution is only counted if a customer confirms it or ends the conversation without asking for more help. This model has grown into a business line with over $100 million in annual recurring revenue, backed by a $1M performance guarantee if targets aren't met. Developer-focused tools like the AI code editor Cursor illustrate a different, more complex path. After initially offering a set number of "fast requests," Cursor switched to a usage-based credit pool. This model directly ties a customer's cost to the underlying expense of the large language models (LLMs) being called, where a simple completion is cheap but a complex, multi-file refactor is expensive. Implementing outcome-based models presents significant hurdles for vendors. Key challenges include the difficulty of defining and reliably measuring business outcomes, the potential for revenue volatility, and the need for a robust data infrastructure to track value transparently. For customers, the primary benefit is the de-risking of software investments—they only pay for tangible results. The move to outcome-based pricing is seen as the next evolution of usage-based models, adding a layer of vendor accountability. While usage-based pricing meters consumption, outcome-based pricing holds the vendor responsible for achieving a specific result. Analysts predict that by 2025, over 40% of B2B SaaS companies will offer some form of outcome-based pricing.