17 Comments
User's avatar
Bocar Dia's avatar

Great analysis as always Kyle! I'm curious about hybrid approaches - have you seen successful examples where companies started with a simpler model (like per agent) but incorporated outcome-based bonuses as training wheels before fully transitioning to the results-based model? This could address the attribution challenges while still moving toward the model you see winning long-term.

Expand full comment
Kyle Poyar's avatar

100% — there are a ton of hybrid models, some folks get very creative!

Expand full comment
Silvia Arrigoni's avatar

Comprehensive article backed by solid research. And a great read. Super helpful. Thank you!

Expand full comment
Kyle Poyar's avatar

🙌

Expand full comment
GregD's avatar

Another incredible research post by Kyle. I appreciate the breakdown and pros and cons of each.

Expand full comment
Kyle Poyar's avatar

🙌

Expand full comment
gsp's avatar

Incredibly helpful to my team and our business model. Huge thanks!

Expand full comment
Kyle Poyar's avatar

Glad to hear it!

Expand full comment
Steven Forth's avatar

This is similar to the Ibbaka Agent Pricing Layer Cake. In many cases these different layers will be combined and we will see hybrid pricing of agents. https://www.ibbaka.com/ibbaka-market-blog/pricing-in-the-agent-economy

Expand full comment
🎲 Monetization Product Manager's avatar

Very nice 👌

Expand full comment
Jorge Bestard's avatar

great read!

Expand full comment
Sarthak Mishra's avatar

Hi Kyle,

How have you seen these models getting sold to large organizations which generally require a cost estimate upfront (monthly or annually) given that the usage drives the overall costs?

Expand full comment
Manny Medina's avatar

the way i have seen this work is by selling buckets of X. X being outcomes or activities. and then agreeing on a price when the customer go over or the vendor eating the overage and revisiting the buckets every 6 months to a year

Expand full comment
Sarthak Mishra's avatar

Thanks @manny for this insight.

Expand full comment
adrian's avatar

Kyle your survey won't allow me to select a company being founded before 2018

Expand full comment
Kyle Poyar's avatar

There should be an option to select “Earlier”

Expand full comment
Al Sargent's avatar

Thank you Kyle! Great to see some fresh thinking around how to price AI services.

In designing pricing and packaging, I try to factor in Explainability (how difficult is it, in practice, for a salesperson to explain the pricing model), "Estimateability" (how hard is it for a customer to determine the quantity they'll need to purchase),l and Traceability (how hard is it for a customer to determine how much they've consumed).

Do you have any thoughts around how each of these models performs in terms of ease of explanation, quantity estimation, and tracking consumption? (Perhaps a follow-up post...)

Expand full comment