Hacker News Reader

Comment by davidjgraph | Hacker News Reader

This doesn't sound like the complete set of metrics that should be measured. Surely, the key thing is how much money each model made? What is the average revenue per user in either case for those users converted?

Later on the average lifetime value of a user in each model is needed (plus the cost to get and convert them).

Yes, conversion is part of the equation, but this feels like a vanity metric on its own.

Just so I'm not sounding pedantic, our SaaS tool has a conversion rate of over 40% because it's completely free, permanently. That's the boundary case of moving the pricing.

View on HN

Just looking at this for one of the first times (new to this part of the game), it seems like this is a step-by-step progression from 'we have users!' to 'users are paying!' If I were working my way forward yours would absolutely be my next question, along with figuring how to target the 20% of customers that drive 80% of my revenue.

Do you think that it would be better to have the revenue model fully thought out (through to revenue/user and lifetime value) before a/b testing the pricing models?

There really is no such thing as "fully thought out" here w.r.t what the customers will actually do with it. If you look again at the article, they went through this kind of progression:

- Best guess at a starting price model

- Revisited price model after poor test performance. Did research, revamped model based on research

- Revisited price model after poor test performance. Simplified plan to match the philosophy of business. Performance greatly improved

It's about testing the ideas, not keeping them in your head. What you think works and what actually works are often counterintuitively different.

I'm OK with optimizing this single step in the process. Things can change so much elsewhere that it's rarely all that prudent to perform a more complete analysis as you suggest, especially for a young, high growth service/company.

Reply on news.ycombinator.com