The usual version of this comparison is a table of monthly prices with Chatwoot at the bottom. It is not a useful table, because the three options do not charge for the same thing. What you are really choosing between is three different cost structures, and which one suits you depends on facts about your team that a price list cannot know.
Here is how to work it out for yourself, including the cases where staying put is the right answer.
Three ways a helpdesk charges you
Per seat
The classic model, and still how most of Zendesk and Intercom’s plans are shaped. You pay monthly for each agent who can log in.
It is predictable, which is genuinely worth something. It also means your bill is driven by headcount rather than by value received – and that is where it starts to bite. The part-time weekend agent costs what the full-time one costs. The warehouse supervisor who answers four messages a month costs a full seat. Seasonal teams pay to scale up and then argue internally about scaling back down.
Teams respond by rationing logins, which is how you end up with three people sharing an account and no idea who answered what.
Per outcome
The newer model, which the market has been moving toward as AI handles more of the first response – charging for resolutions or conversations handled rather than for seats.
It sounds fairer, and for a small team it often is. But notice what it indexes on: your bill now grows with your customer volume. A good quarter costs you more. A product launch costs you more. The cost you cannot forecast is the one that shows up in a board meeting.
Worth modelling at three times your current volume, not your current one.
Infrastructure plus ownership
Self-hosted Chatwoot. You pay for a server, and you pay for someone to own it. Both are roughly flat in agent count and in conversation volume, until you outgrow the server and buy a bigger one.
This is the structure that behaves differently from the other two: adding the tenth agent, or the ten-thousandth conversation, does not change the bill. That is the entire argument, and it is also why the argument is weak for small teams – there is nothing to flatten yet.
What self-hosting actually costs
Any honest comparison has to put real numbers on this side too. Four line items:
- The server. Chatwoot needs the app, PostgreSQL, Redis and a worker. Modest for a small team, and it grows with conversation history and attachments rather than with agents.
- Setup. One-off, but real – deployment, TLS, outbound email, channels, and importing what you are migrating from.
- Maintenance. Upgrades, backups, monitoring, the occasional incident. A few hours a month if nothing breaks. Price those hours at what the person is actually worth.
- Being on call. The one people leave out. If the channel going down matters, somebody has to be reachable, and that has a cost whether or not anything happens.
Add those honestly and self-hosting is still usually cheaper past a certain size. Add them dishonestly and you will make a decision you resent in six months. We went through this in more detail in self-hosted Chatwoot vs Chatwoot Cloud.
What you give up
Chatwoot is a strong multi-channel inbox. The incumbents are platforms, and they have had far longer to build out the edges.
Expect to find gaps in the mature end of the feature set: deep analytics and forecasting, the breadth of their app marketplace, product tours and in-product onboarding, and the most polished AI tooling. If your team genuinely uses those, the subscription is buying something real and the cost comparison is not the whole picture.
The honest test: open your current tool and list the features your team touched this month. Not the ones that sold you on it – the ones in daily use. That list is usually shorter than people expect, and it is the only list that matters.
The migration cost nobody budgets
Switching is not free even when the destination is cheaper:
- History. Decide early whether you are migrating past conversations or archiving them somewhere read-only. Both are defensible; drifting into neither is not.
- Integrations. Anything wired into your CRM, order system or billing has to be rebuilt. Count them before you commit, because there are usually more than anyone remembers.
- Automations and macros. Years of accumulated rules and canned replies. Most of them are dead weight, but finding out which takes someone’s attention.
- Retraining. A fortnight of slower handling while agents learn a new interface, plus the complaints that come with it.
- Running both. You will overlap for a period and pay twice during it. Plan the overlap rather than discovering it.
When the switch clearly pays
- Your agent count is the problem. Per-seat pricing and a large or seasonal team is the textbook case. The saving is immediate and it compounds.
- You are WhatsApp-first. If WhatsApp is your main channel rather than a bolt-on, Chatwoot handles it as a first-class citizen and you are not paying for a suite built around email ticketing.
- You are reselling it. Per-seat pricing does not survive contact with being the one who charges. See multi-tenant Chatwoot.
- Data residency is a requirement. Then this is not a cost decision at all, and the cost saving is a bonus.
- You want the database. Direct SQL access changes what reporting you can build. No subscription tier sells you that.
When it does not
Three or four agents, no infrastructure person, and a tool your team likes: stay. The saving is smaller than the disruption, and “cheaper” is not a strategy when the number is small in absolute terms.
Likewise if your workflow genuinely leans on the incumbent’s ecosystem – deep CRM integration, their automation, their reporting feeding something else. Replacing a platform with an inbox plus a list of things you will build later is how a cost saving turns into a project.
How to actually decide
Build a three-year model with your own numbers. Not one year – the migration cost lands in year one and distorts it.
On one side: your current subscription at the headcount and volume you expect to reach, not today’s. On the other: server, setup, maintenance hours priced honestly, on-call, and the migration. Then ask whether the difference is large enough to be worth a quarter of somebody’s attention.
Sometimes it obviously is. Sometimes the answer is to stop thinking about it and get back to answering customers, which is a perfectly good outcome for an afternoon’s work with a spreadsheet.
We deploy and maintain self-hosted Chatwoot, including migrations from hosted helpdesks and WhatsApp Cloud API integration. If you have run the numbers and want the saving without owning the server, get in touch. If you have run them and the answer is to stay where you are, we will tell you that too.