Outcome-Based Pricing: The Shift

The Shift

• Salesforce first mover: pricing AgentForce on customer revenue generated, not tokens.
• OpenAI following: largest customers pay only when AI completes the job.
• Framing: token sellers sell compute; outcome sellers sell labor.
• Analogy: fixed-price vs. time-and-materials contracts — effectively a performance guarantee.

Historical Precedent

• Siebel Systems predates Salesforce on this model.
  • CRM at $50/year didn't work well; wrapped in “total success” it became $20–30K/year.
  • Price up 1,000x, but customers happier because they got the full solution.
• OpenAI's strategic pivot: consumer/subscription focus let Anthropic dominate enterprise.
  • Outcome pricing as leapfrog: bypass token-based enterprise deals with 10,000x higher price points.
  • Example: drug discovery worth hundreds of billions — charge 10% of that.

The CPM/CPC/CPA Analogy for AI

Digital ads pricing tiers map directly to AI pricing:

Prediction: a pricing picker emerges, like Google/Facebook Ads — users choose: pay by flops, tokens, or outcomes.

From OpenAI's side: outcome pricing creates a price discovery mechanism. Customers reveal which tasks are worth $10K vs. $1M per unit token — helps OpenAI catch up on enterprise after ceding ground to Anthropic.

Societal Unlock: Banks and Job Preservation

• Regional bank example: spending $2/million tokens = trivial revenue, no priority from OpenAI, no talent, nothing gets done.
• New model: OpenAI delivers “3x customers at half the price” and takes a large share of the lift.
  • Sam now cares because the price point is thousands of times larger.
  • Bank actually achieves the outcome, keeps its people.

Next Steps