pricing philosophy
Alas, we've go to charge something - so how do we think about this?
We should have "one price" for individual users.
- Charge per seat - it's not a team decision, it's an individual decision as this product is so individually focused
- It should be free for a small amount of usage - it should roughly mirror what you get free in other ai first editors
- It should be fixed monthly, in buckets like $20/60/200 - we can probably mirror what other ai editors do just so it's familiar
- Our costs come from two things (1) coding costs and (2) PostHog usage. We need to adjust upwards the bucket someone is in once they go over a certain amount of cost.
- On coding we should make some but not a ton of margin (as it's super competitive), just enough extra because it's automatic whilst you sleep. Say 35% to start with. We can reduce over time if needed.
- On PostHog costs, we should just make these whatever they would pay as a customer of those products, else there's a weird incentive to not use PostHog through Twig.
- In future we can think about clever ways to reduce pricing (for example make PostHog cheaper because of the revenue from coding or vice versa). At the moment we need to start with something reasonable. We can easily go cheaper, it'll be hard to go more expensive.
- If a team of say 3 are using it, the infra cost should be split across all of them. We need to make clear in the UX that extra seats get cheaper because of this. I can't see a simpler way to handle this (but would love one).
- We should make available in the docs pages a very comprehensive copy of our PostHog pricing as just one huge table, and on the pricing page we just show a handful of plans (and we link out to our cost structure if people need to go even deeper).
- The product generally is try it and see, with transparency. We should avoid at all costs that people feel like they need to read a huge table of prices of all our existing PostHog prices.
- This is a mass market product so shouldn't be exclusive feeling. We want to equip every developer, not the Fortune 500. The good news is that others are making a ton of revenue without being prohibitively expensive because the value is there for any engineer as AI is so powerful.
A few things to avoid from others:
- A set fee without pure cost driving the bucket changes causes bad incentives - for example, changing to a crappy model to save money but worsening the experience without letting the user choose.
Some things that are different from PostHog:
- We aren't aiming to be the cheapest. No one else can provide what we provide. More revenue means we can just reinvest more into engineering to extend what this does.
Todo: figure these out
- Can users bring their own API keys? How does this affect pricing if so?
teams and enterprise plans
We will need these, but we've not built a lot of collaborative features yet.
In future we'll want to build a triage page, permissions and all sorts.
This will likely be a simple increase in the per user cost for teams functionality, and perhaps a flat fee for enterprise that is more custom. We will cross that bridge when we come to it.