Per-seat pricing is the most successful billing model in software, and it deserves to be: it's easy to understand, it scales revenue with customer size, and at the point of purchase it always looks reasonable. Ten seats at a low monthly figure is a rounding error, and that's exactly the number you evaluate.
The problem isn't that it's expensive. It's where the expense comes from — and what a per-head multiplier does to how a company behaves once it's in place.
The mechanism, in one sentence
Consider what a chat platform actually spends per user: some database rows, some storage, a websocket connection, a bit of search index. Now consider a 200-person company where forty people are heavy users, sixty are occasional, and a hundred check in twice a day. The infrastructure cost is dominated by the first group. The invoice is dominated by all two hundred.
That gap is not a scandal — it's how software companies fund product development, support and the security team you're glad exists. But it does mean the price you pay has less and less to do with the service you consume as you grow, and it's worth knowing which of those you're buying.
The compounding shape
Three multipliers stack, and each is individually reasonable:
- 1Headcount growth. You hire. The bill grows proportionally, whether or not the new people generate proportional value from the tool.
- 2Tier upgrades. You need one feature — SSO, an audit log, a retention policy — and the upgrade re-prices every seat, not just the seats that use it. This is the big one, and it's the step that surprises people.
- 3Tool count. Chat, docs, tickets, CI, design, CRM. Each is individually defensible per seat. Together they can add up to a per-employee software cost that would startle you if anyone ever added it up. Try adding it up.
| 10 people | 50 people | 100 people | |
|---|---|---|---|
| Base plan at $10/seat/mo | $1,200/yr | $6,000/yr | $12,000/yr |
| After a tier jump to $15 | $1,800/yr | $9,000/yr | $18,000/yr |
| Six tools averaging $12/seat | $8,640/yr | $43,200/yr | $86,400/yr |
| Flat-cost self-hosted chat | ~$150/yr | ~$600/yr | ~$600/yr |
Illustrative figures, not quotes from any specific vendor. The row that matters is the shape difference: three of these lines are functions of headcount and one is a function of load.
The behaviours it quietly encourages
This is the part that gets less attention than the arithmetic, and it's arguably more damaging.
Hesitating before adding someone
The contractor who'd benefit from being in #engineering for two weeks doesn't get added, because adding them is a purchase. Small friction, applied constantly, in the exact place where you want zero friction.
Seat archaeology
Somebody's job becomes auditing who's still using what. This is genuinely necessary under per-seat pricing and produces nothing of value — it's pure overhead created by the billing model.
Shadow accounts
Shared logins, a team account passed around, people working around the seat count. The security implications are exactly as bad as they sound, and the cause is a pricing model rather than carelessness.
Tier-driven architecture
The clearest sign it's shaping decisions: when you find yourself choosing a technical approach to avoid needing a feature that lives one tier up. That's the pricing model designing your systems, which is not what you hired anyone for.
When per-seat is genuinely the right deal
It would be daft to argue the model is always wrong. Per-seat pricing is a good buy when:
- You're small and staying small. Ten seats forever is cheap, and the alternative costs hours you don't have.
- You need the support relationship. A vendor with an SLA and a human on the other end is worth real money, and per-seat is how that gets funded.
- Compliance requires a vendor. Sometimes the auditable third party is the product.
- The tool is genuinely differentiated. If nothing else does the job, the pricing model is a detail.
The problem is that team chat fits none of those especially well. Chat is a commodity in the good sense — the category is mature, the feature sets have converged, and several excellent open source implementations exist. It's the last place where a per-head multiplier earns its keep.
What flat cost changes
Move to a tool priced on infrastructure — or not priced at all — and a set of small decisions stop being decisions:
- Adding a contractor to a channel is adding them to a channel.
- Nobody audits seats, because there are no seats.
- The feature you need isn't in a tier, because there are no tiers.
- Hiring plans and software budgets stop being coupled.
- Renewal season loses one conversation.
The cost, stated fairly: with self-hosted software you're buying hours instead of seats, and hours are more expensive than most people account for. With free hosted software you're accepting a narrower feature set — Fluid, for example, has no huddles, no voice or video, no SSO/SAML and no native mobile apps. Neither of those is free lunch; they're a different lunch.
No seats to count. No tiers to jump.
Fluid Chat is $0 whether you self-host it or use the hosted version, for unlimited people, with every admin feature included.
The test worth running
Take your per-seat tools, multiply each by your headcount and by twelve, and add them up. Then look at that annual figure and ask, tool by tool, whether the value scales the way the bill does.
For your CRM it probably does — more salespeople genuinely means more value from a CRM. For the place where people say good morning and paste stack traces, the honest answer is usually no.