Salesforce bets on outcome pricing as AI upends 25-year SaaS model

Salesforce is letting customers pay for its Agentforce AI product based on revenue gains or cost savings, a shift that could upend the per-seat subscription model it pioneered 25 years ago.

"Customers want to buy and price in different ways, something I've come to understand deeply recently," Chief Executive Officer Marc Benioff said on the company's earnings call.

The flexible pricing, which mirrors Palantir's customized contracts blending fixed fees with usage- and outcome-based charges, has helped Salesforce close "very large deals," Benioff said. The shift follows a 22.6 percent surge in Salesforce shares on Aug. 27 after the company reported fiscal second-quarter results that beat estimates, with revenue up 11 percent to $11.3 billion and Agentforce annual recurring revenue passing $1.5 billion, up more than 240 percent.

The experiment will determine whether legacy enterprise software companies can reinvent themselves as AI agents from Anthropic and OpenAI reduce how often employees interact directly with applications. Salesforce also launched Claudeforce last week, a service letting customers use Anthropic's Claude to complete tasks inside Salesforce apps, with plans to charge for every third-party AI call to its data.

Per-seat pricing loses its footing

Twenty-five years ago, Salesforce led the industry's shift from one-time license fees to per-employee subscriptions, lowering upfront costs for small businesses and shifting upgrade and maintenance burdens to vendors. That model fueled two decades of SaaS growth. AI inverts the logic: as enterprises route complex tasks through Claude and other agents, employees interact with applications less often, eroding the foundation of per-seat billing.

Benioff acknowledged the uncertainty, saying Salesforce is following startups rather than leading the change. OpenAI has begun offering some large customers the option to pay only when an AI completes a task. Customer-management startups Sierra and Fin — the latter being acquired by Salesforce for $3.6 billion — charge only after tasks are finished. Coding assistant Cognition promises customers up to $10 million in credits if it fails to deliver engineering results worth the payment.

Attribution disputes loom

Outcome-based pricing carries execution risk. Payment processor Stripe has warned that sales conversions "may result from product changes, marketing campaigns, or seasonal factors" rather than the software itself. "Unless attribution rules are clear, customers may dispute whether results should be credited to the software vendor," Stripe said.

The concern has precedent. Splunk, the software monitoring company, saw revenue decline during its transition from license to subscription pricing. Analysts including Guggenheim's John DiFucci said Salesforce's latest bookings, retention, and AI product data eased fears of a "SaaS apocalypse," while Evercore ISI's Kirk Materne said Claudeforce shows AI models and enterprise systems of record can co-evolve.

Salesforce shares, trading at about 16 times forward earnings, have climbed roughly 23 percent since the earnings report. The outcome-based pricing push, if it works, could let Salesforce capture a share of the value its AI agents create — Benioff framed the goal as charging $2 for every $20 or $40 in revenue the software helps generate. The risk is that attribution disputes slow deal flow, as they did for Splunk, and that the shift to usage-based billing compresses near-term revenue recognition.

This article is for informational purposes only and does not constitute investment advice.