Why this matters now
For fifteen years the default software price was a seat. In 2026 it is a unit of work. Usage-based pricing has reached 38 per cent adoption across SaaS, up from 27 per cent in 2023, and 61 per cent of vendors now run a hybrid model that pairs a base subscription with a metered layer. Pure per-seat pricing, where headcount is the only value metric, has shrunk to roughly 8 per cent of the market, and IDC forecasts that 70 per cent of software vendors will refactor away from it by 2028.
Two forces did this. AI and compute-heavy features made flat plans dangerous, because a flat price hides a variable cost and the margin absorbs the difference. And buyers pushed back on paying per seat for software that increasingly does the work rather than helping a person do it. Credit-based pricing surged 126 per cent year on year as the fastest way to monetise AI features alongside existing plans, though most teams describe credits as a workaround rather than an answer, and the market is already swinging back toward predictability.
This matters at the blueprinting stage rather than at the pricing meeting. A go-to-market blueprint that assumes a fixed price per customer produces a CAC model that stops being true the moment consumption varies. The same change has already arrived on the buying side of a startup's own stack: HubSpot prices Breeze agents at roughly US$0.50 per resolved conversation, Salesforce sells Agentforce at US$2 per conversation, and Gartner found 56 per cent of marketing leaders increasing the share of martech budget on consumption pricing.