We deleted per-seat pricing

Groove charged per seat for 12 years. Every help desk does. It's such a settled question in the category that I don't think I seriously reconsidered it once in a decade.

Helply charges $0 per seat and $0.50 per resolved ticket. The platform is free. The seats are free. You pay when the system does the work.

Everyone's first reaction is that it's a marketing position. It started as a math problem.

The math that forced it

If your AI works, the customer needs fewer support people. Fewer people means fewer seats. Fewer seats means you get paid less for a product that just got better.

That's the whole thing. Per seat pricing means your revenue goes down exactly in proportion to how well your product performs.

The industry's answer has been to keep the seats and add a per resolution fee on top of them. So the customer pays for the humans and pays again for the AI replacing the humans. I understand why that happened, because it protects the existing revenue line while you figure things out. It also means the buyer is now doing arithmetic on your invoice that ends with them looking for an alternative.

We didn't have an existing revenue line to protect on the new product, which is the only reason we could do the clean version. Groove was funding us. That's a real advantage and I want to name it rather than pretend this was purely courage.

What broke first: forecasting

Per seat revenue is boring and beautiful. A customer signs for 12 seats, you know what December looks like.

Usage revenue moves. A customer's ticket volume spikes because they shipped a bug. It drops because they fixed their onboarding. It drops because we worked, and their deflection got better, which is the good outcome and also a smaller invoice.

Our first 2 months of forecasts were nonsense. Not slightly off. Nonsense. We were modeling on signed accounts and the accounts weren't the unit anymore.

What we do now is model volume bands per customer with a floor and a ceiling, look at the aggregate, and accept a wider error range than I'd like. Net revenue retention is a much better number to watch than new logos, which took me longer to internalize than it should have given that I've been doing this for 20 years.

What broke second: comp

You cannot pay a sales team on contract value when there isn't one.

We spent a while getting this wrong. Paying on estimated first-year usage creates an obvious incentive to over-estimate, and the estimate is a guess that the rep controls. Paying on realized usage means a rep closes a deal in March and finds out what they earned in September, which is not a compensation plan, it's a lottery.

Where we landed is a mix, weighted to the first 90 days of actual usage, with a smaller bonus at 12 months for accounts that grew. Not elegant. It works, mostly, and I expect to change it again.

What broke third: the first invoice conversation

Usage pricing means the customer doesn't fully know what they'll pay until they've paid it. That's uncomfortable for a finance team and it should be.

Two things fixed most of it.

Exclusions we can defend. We only bill for tickets where something actually got done. Spam doesn't count. Notifications don't count. No-touch tickets don't count. That list of exclusions was harder to design than the prices themselves and it's the reason the pricing page survives a hostile read, because the first question every buyer asks is "so you're incentivized to resolve garbage."

And a spend cap. Any customer can set one. Almost nobody hits it, but the existence of it moves the deal, because the fear was never the price, it was the unbounded-ness.

Was it worth it

Yeah. It's the single clearest thing about us in a category where everyone's messaging sounds the same. $0 per seat is a sentence a buyer repeats to their boss correctly, which is not something you can say about most positioning.

It also does the qualifying. Teams who want to pay per seat because it's predictable self-select out, and that's fine.

The unresolved part is margin. Our cost of serving a resolution moves with model pricing, and model pricing has gone one direction so far. If that reverses, or if a customer's mix shifts hard toward the expensive kind of ticket, we find out fast and in public. We watch it weekly. I've got a spreadsheet I don't love.

$0.50 is right today. I'm not going to pretend I know it's right in 2028.