It's something lots of people talk about but few succeed at. I'd like to argue it's partially because a confusion about where to solve it.
The attached Wardley map explain this better visually, but in essence:
- Defining "positive outcome" is hard and unless solved the rest is ofc pointless. This is highly contextual.
- Correlating the usage events with the outcome events (and/or queries) in time and by identity must be done BEFORE pricing is done. Pricing systems aren't built to do this.
- Once the combined event exists, existing usage-based pricing systems can be used without modification.
Makes sense? (Edit: Improved wardley map)