Chatwoot vs Crisp vs Intercom vs Plain: Customer Support Software 2026

Picking a customer support platform used to be a checkbox exercise: install a chat widget, point it at a shared inbox, done. That stopped being true around 2024, and by 2026 the decision has real money and real architecture attached to it. AI agents now resolve a chunk of tickets automatically, pricing models have splintered into per-seat, per-workspace, and per-resolution camps, and the gap between an open-source tool you host yourself and a polished SaaS you rent has never been wider.

I run a small software studio, Warung Digital Teknologi, and across 50+ client projects over the last 11 years I have shipped a Helpdesk Ticketing system from scratch, a ServiceBot AI Helpdesk on top of the OpenAI API, and integrated third-party support widgets into e-commerce and SaaS products more times than I can count. So I have sat on both sides of this fence — building the inbox, and buying it. This comparison covers the four tools I keep coming back to in client conversations: Chatwoot, Crisp, Intercom, and Plain. They sit at four genuinely different points on the cost-versus-control curve, which is exactly why putting them side by side is useful.

The four tools at a glance

Before the deep dive, here is the short version of who each tool is actually for. These are not interchangeable products that happen to compete — they are built around four different philosophies.

• Chatwoot — open-source (MIT-licensed), self-host for zero license cost or pay for cloud. The choice when you want full control and have engineers to run it.
• Crisp — per-workspace pricing, not per-seat. The choice when you have a growing team and refuse to pay per head.
• Intercom — the premium incumbent with the most mature AI agent (Fin). The choice when budget is not the constraint and resolution quality is.
• Plain — a developer-first, API-native support platform for B2B teams that live in Slack, Linear, and their own code. The choice when "configure a helpdesk" is the wrong shape for your workflow.

Pricing in 2026: the part everyone gets wrong

Pricing is where these four tools diverge hardest, and where the headline number on the marketing page hides the real total. Here is the comparison I wish someone had handed me before my last vendor evaluation.

Three things in that table deserve a closer look, because they are the data points that actually change the decision.

1. Crisp's per-workspace model is the quiet cost killer. Crisp's Essentials tier at roughly €95/month includes up to 10 seats at no extra charge, and the Plus tier (~€295/month) covers up to 20. Compare that to per-seat tools: for a 10-person support team, Crisp's structure can cut your bill by 50–67% versus Intercom or Zendesk. When I scoped support tooling for a client running a 12-agent operation, the per-seat quotes came in 3–4x higher than Crisp's flat workspace fee. If your team is bigger than your budget, this is the single biggest lever.

2. Intercom's $0.99-per-resolution Fin pricing is honest but spiky. Fin charges $0.99 per "outcome" — a resolution where the customer confirms they are helped or exits without escalating. You are billed once per conversation even if Fin answers five questions, which is fair. But there is a mandatory minimum of 50 resolutions per month, no volume discounts, and no spending cap. Intercom's own published case studies put real-world Fin resolution rates at 42–50%, so if you handle 5,000 conversations a month, budget for roughly 2,100–2,500 resolutions — about $2,000–$2,500 in Fin charges alone, on top of $39–$99 per seat. The model is transparent; the bill is not small.

3. Chatwoot's "free" is not free — but it is cheap. Self-hosting Chatwoot costs nothing in license fees, and that headline is real. The catch is the operational tail: WhatsApp and SMS delivery fees pass through regardless, you own backups, monitoring, and upgrades, and engineering time is not free. In my experience running a self-hosted Chatwoot instance for an internal team, the actual monthly out-of-pocket was a $12 VPS plus WhatsApp Business API message fees — but the first weekend of setup cost me real hours. The trade you are making is dollars for engineering time, and that only pays off if engineering time is the thing you have a surplus of.

Chatwoot: full control, if you can run it

Chatwoot is the open-source omnichannel support platform — live chat, email, WhatsApp, Telegram, Instagram, and an API, all under an MIT license you can self-host. For a studio like mine, that license matters: I can fork it, read the Rails source, and deploy it on the same Hostinger VPS that runs everything else, with no vendor lock-in.

What I like about it in practice: the data stays on infrastructure I control, which is non-negotiable for some clients in regulated spaces. The agent inbox is genuinely good — keyboard-driven, fast, with canned responses and labels that map cleanly to how a real support queue works. The WhatsApp and Telegram channels are first-class, not bolted-on, which matters enormously for the Indonesian market most of my clients serve.

Where it bites: you are the ops team. Upgrades, Redis and Postgres tuning, email deliverability, and uptime are all on you. Captain — Chatwoot's AI layer — exists, but it is newer and less battle-tested than Intercom's Fin. If your team cannot comfortably run a Dockerized Rails app with Postgres and Redis behind a reverse proxy, the "free" self-hosted path will cost you more in incidents than a SaaS subscription ever would.

My take: Chatwoot is the right call when data ownership or per-seat cost at scale is the dominant concern, and you have at least one engineer who treats the instance as a real service. For a 5–10 person team that already runs its own infrastructure, the cloud plan at $19–$39/month is also a legitimately good deal without the ops burden.

Crisp: the team-scaling cost play

Crisp is the one I recommend most often to non-technical founders, and the reason is almost entirely the pricing model. Per-workspace billing means a 10-person team on the Essentials plan pays one flat fee, not ten. For small businesses that grow their support headcount faster than their revenue, that structure removes the "every new hire raises the support bill" tax that per-seat tools impose.

Feature-wise, Crisp covers the essentials well: shared inbox, chatbot, a knowledge base, a CRM, co-browsing, and campaigns, all in a clean UI. The free tier (2 agents) is genuinely usable for a side project or early-stage product, which is more than I can say for most "free" plans that are really just trials in disguise. When I dropped Crisp into a client's e-commerce storefront, the widget was live and answering questions inside an afternoon.

The limits show up at the top end. Crisp is not built for deep, programmable, B2B support workflows — its AI and automation are competent but not best-in-class, and very large operations will eventually outgrow the workspace-tier seat caps. It is a sweet-spot tool: excellent from 2 to ~20 agents, less compelling above that.

My take: If cost-per-agent-at-scale is your pain and you want something live the same day, Crisp is the highest-value pick in this group. I would reach for it over Intercom for any SMB that is not running an AI-heavy support operation.

Intercom: premium, and the AI actually works

Intercom is the incumbent everyone compares against, and in 2026 its differentiator is Fin, the most mature AI support agent on the market. The per-resolution pricing I broke down above is the honest version of "pay for outcomes," and Fin's 42–50% real-world resolution rate is, frankly, good — it is the difference between deflecting half your inbound volume and not.

Having integrated AI support myself when I built ServiceBot AI Helpdesk on the OpenAI API, I have respect for how hard that resolution quality is to hit. Getting an AI to confidently answer half of real customer questions without hallucinating into a refund-policy lawsuit is non-trivial. Intercom has spent years and a lot of money on the retrieval and guardrail plumbing, and it shows. The wider product — Copilot for agents, proactive messaging, reporting — is polished and deep.

The cost is the cost. Five agents on the Advanced plan is $495/month in seats before a single Fin resolution is billed, and the add-ons (Copilot at $29–$35/seat, Pro analytics at $99/month, Proactive Support Plus at $99/month) stack quickly. Intercom is a premium product priced like one. For a funded SaaS with high ticket volume where deflection pays for itself, the math works. For a bootstrapped studio, it usually does not.

My take: Choose Intercom when AI resolution quality is the metric that matters and you can afford to pay for it. If you are counting dollars, the same budget buys you Crisp plus a part-time human, and that combination beats Intercom for a lot of smaller operations.

Plain: support built like infrastructure, for developers

Plain is the odd one out, and the most interesting. It is a B2B support platform built on the premise that engineering-heavy teams do not want a helpdesk to configure — they want an API to build on. The framing they use is apt: Plain is to support what Stripe and Twilio are to payments and messaging. You work from Slack, Linear, or your own tools, and Plain is the programmable layer underneath, not a separate UI your team has to live in.

For a developer audience this resonates hard. If your support tickets need to create Linear issues, your customer context lives in your own Postgres, and your engineers triage from Slack anyway, a traditional helpdesk is friction. Plain's API-first model lets you wire support into the workflow you already have instead of bending your workflow around a vendor's inbox. I have built exactly this kind of glue by hand for clients, and a platform that ships it natively saves real time.

The cost reflects the audience. Foundation is $39/month for one seat, Expansion jumps to $299/month for three, and Frontier is $1,055/month for six. Per-seat pricing that climbs steeply means Plain is best for small, senior, expensive support teams — not for scaling a large pool of frontline agents. It is also genuinely not for non-technical buyers: the value only shows up if you are willing to build on the API.

My take: Plain is the pick for B2B SaaS with an engineering-led support culture, where customers are other developers and "support" means a senior person solving a hard problem, not a queue of password resets. For that specific shape of team, nothing else here fits as cleanly. For everyone else, it is overkill.

Decision matrix: which one for which team

The one thing every comparison skips: total cost of ownership

Every vendor page sells you the monthly subscription. Almost none of them tell you the real total cost of ownership, which is where I have watched budgets go sideways. Three line items get forgotten:

• Channel delivery fees. WhatsApp Business API and SMS are billed per message by the carrier, no matter which platform fronts them. On a self-hosted Chatwoot or a cloud tool alike, a high-volume WhatsApp support line can cost more in delivery than in software.
• AI usage on top of seats. Intercom's Fin is the clearest example — $0.99 per resolution with no cap means your "support bill" has a variable component that scales with success. That is fine if you model it; it is a nasty surprise if you do not.
• Engineering and migration time. Self-hosting trades dollars for hours, and migrating between platforms later is expensive in both. Choosing wrong now compounds.

The lesson from my own projects: model 12 months of total cost — software plus channel fees plus AI usage plus the engineering hours to run it — before you sign anything. The cheapest sticker price is rarely the cheapest year.

Frequently asked questions

Is self-hosted Chatwoot actually cheaper than paying for Crisp or Intercom?

In raw dollars, yes — you can run it for the cost of a small VPS. But only if you already have the engineering capacity to operate it. Factor in setup, upgrades, monitoring, and incident response, and for a team without spare engineering hours, a cloud subscription is usually cheaper in true cost. Self-hosting wins when engineering time is your surplus resource, not your bottleneck.

What does Intercom's Fin actually cost per month?

Fin is $0.99 per resolution with a 50-resolution monthly minimum and no cap. At a 42–50% resolution rate, a team handling 5,000 conversations a month would see roughly 2,100–2,500 resolutions, or about $2,000–$2,500 in Fin charges — separate from the $39–$99 per-seat fees. Always model it against your actual conversation volume.

Why is Crisp cheaper for larger teams?

Crisp bills per workspace, not per seat. Its Essentials plan (~€95/month) includes up to 10 agents and the Plus plan (~€295/month) up to 20, all for one flat fee. Per-seat competitors charge for every agent, so for a 10–20 person team Crisp can be 50–67% cheaper.

Who should choose Plain over the others?

B2B SaaS teams whose support is engineering-led — customers are developers, the team works from Slack and Linear, and tickets need to flow into the company's own systems through an API. Plain's per-seat pricing climbs steeply, so it suits small senior teams, not large frontline pools.

Can I start free and upgrade later?

Yes. Crisp's free tier (2 agents) and Tawk.to (free forever, paid white-label removal from $19/month) are both genuinely usable starting points, and Chatwoot's self-hosted edition is license-free. The catch is migration cost — moving conversation history and integrations between platforms later is real work, so pick a tool you can grow into rather than one you will outgrow in six months.

Bottom line

There is no single winner here, and any comparison that crowns one is selling something. After building support systems and buying them across more than a decade of client work, this is the framing I trust: Chatwoot if you want control and have the engineers to earn it, Crisp if you want the best cost-per-agent at scale with zero setup pain, Intercom if AI resolution quality justifies a premium bill, and Plain if your support is really an engineering workflow wearing a helpdesk costume. Model the full 12-month total cost — seats, channel fees, AI usage, and ops time — and the right answer for your team usually becomes obvious.