Why we charge for customers, not seats
Seats cost us nothing to serve. AI replies and email do. So that's what the price follows.
Every feedback and support tool we looked at charges per person on your team. It sounds fair until you run the numbers. Support needs a login. Product needs a login. The engineer who fixes the bug should see the thread. The founder wants to read the NPS comments. Five people on Featurebase Professional is $295 a month; on Intercom Advanced it's $425.
So teams start rationing logins, and the whole point of the tool — everyone hearing the customer — quietly stops working.
What actually costs money
We looked at what a workspace costs us to run, line by line.
- A teammate costs close to nothing. A row in a table and a session cookie.
- An AI answer costs about one to two cents: the agent reads a few help-center passages and writes up to a short paragraph.
- An email costs a fraction of a cent. Fine one at a time, real money when a changelog goes to 25,000 customers every week.
- A customer on record costs storage, search indexing and the identify calls that keep them current.
Per-seat pricing charges for the one thing on that list that's free.
So the price follows the rest
Every plan has unlimited seats. What changes between plans is how many customers you've identified, how many conversations the AI closes for you, and how much bulk email you send.
| Free | Pro | Team | Scale | |
|---|---|---|---|---|
| Price, billed yearly | $0 | $32/mo | $99/mo | $249/mo |
| Identified customers | 250 | 2,500 | 10,000 | 50,000 |
| AI resolutions a month | your own key | 100 | 500 | 2,000 |
| Changelog and campaign email | 1,000 | 10,000 | 50,000 | 150,000 |
Past the allowance, an AI resolution is $0.29 and email is $1 per 1,000. Customers are a soft limit: nothing stops, owners get an email at 80% and 100%, and the next plan up is the honest fit. If you bring your own Anthropic key, AI isn't metered at all; you pay Anthropic directly and we add nothing.
Why not just make it cheaper?
We tried. Our first prices were $29, $79 and $199 with a thousand AI replies thrown into the cheapest paid plan. The math was simple: one busy workspace on Pro would have cost more to run than it paid. A tool that loses money on its best customers either raises prices on them later or disappears. Neither is a good thing to sell someone.
The prices now keep every plan comfortably above cost even when a workspace uses everything it's allowed, and the allowances are set well above what we see teams use in a normal month.
What this means for you
Invite everyone. Your support lead, your PM, the two engineers who own the export pipeline, the founder who reads every detractor comment. None of them changes your bill. The only things that do are the ones that grow when your product does.
See the full breakdown on the pricing page.