Is Self-Hosting Chatwoot Actually Cheaper? We Checked the Math

chatwoot, self-hosted, pricing, vendor-evaluation

Is self-hosting Chatwoot actually cheaper than paying for a managed helpdesk? The honest answer: the server itself is not expensive. A VPS meeting Chatwoot's own documented minimum — 4 vCPU, 8GB RAM — runs about $9–12 a month at 2026 prices. What gets expensive is the time someone spends keeping it patched, backed up, and awake at 3 a.m. That distinction is exactly what a widely-cited "true cost of self-hosting Chatwoot" figure blurs together, and it's worth pulling apart before you believe it.

Where the $8,000-a-year number comes from

Search "chatwoot self hosted true cost" and one figure keeps surfacing: self-hosting Chatwoot supposedly runs $8,000 to $24,000 a year once you count everything. It traces to a single source: a May 21, 2026 comparison post on Conferbot's blog, a managed AI chatbot vendor that prices its own product at $3,588 a year flat. Conferbot's own framing does the multiplication for you — self-hosted Chatwoot ends up several times the cost of a Conferbot subscription. That's a normal thing for a vendor comparison to do; it's also exactly why the arithmetic underneath deserves a second look before it gets repeated as a fact rather than as one company's sales argument.

The post isn't wrong about the categories. It lists server infrastructure ("$100–$500+/month for production-grade hosting"), DevOps time ("4–8 hours/month" for deployment, updates, scaling, monitoring), plus unstated costs for SSL, patching, backups, and incident response. Add those up at a contractor's hourly rate and you land in the thousands. What the post doesn't do is check its own first line item against what Chatwoot's documentation actually asks for.

What Chatwoot's docs actually require

Chatwoot's system requirements page is specific, not a vague "production-grade" gesture: 4 CPU cores as the recommended minimum (handles up to 10,000 conversations a day; 8 cores gets you to 20,000), 8GB RAM minimum with 16GB recommended, 60GB SSD, PostgreSQL 14+, Redis 7+, Ubuntu 22.04 LTS or similar. That's checkable against an actual VPS bill.

A 4 vCPU / 8GB Hetzner instance costs about $110 a year, not $1,200–$6,000. Double it for a managed Postgres add-on and off-box backups and you're still nowhere near $100 a month, let alone $500. The "$100–$500+/month" line in Conferbot's post isn't sourced to any vendor or configuration — it's the one number in the comparison you can't check against a receipt.

The 60GB SSD Chatwoot recommends for the main install doesn't change that math either: at typical NVMe block-storage pricing that's a few dollars a month, and the same docs put Redis's own storage need at "100MB, scale up as required" — small enough that it rides along on the same disk rather than needing a separate line item. None of the documented requirements imply a second server, a managed database service, or a CDN; Chatwoot's own deployment guide lists Docker as one of its supported install paths, and a single box running it covers what's specified.

A real-world data point

One measured deployment beats a category estimate. A contractor who runs Chatwoot for clients published a breakdown of an installation handling 348,703 messages across 20 inboxes: 1.8GB of RAM total across the Rails app, Postgres, Sidekiq, and Redis containers combined. Their conclusion was that a 4GB VPS runs this without strain, and that the footprint tracks message volume, not seat count — the opposite of how SaaS helpdesks usually price.

That doesn't contradict Chatwoot's own 8GB recommendation; the vendor's number includes headroom for the 10,000-conversations-a-day tier, and 1.8GB was one team's footprint at lower volume. It does confirm the shape of the finding: nothing here costs $100 a month in hardware, whichever end of the range you land on.

The one line item that's real: Chatwoot's own paid tiers

Here's where a four-figure number does show up honestly. Chatwoot sells two self-hosted upgrades on top of the free Community Edition: Premium Support at $19/agent/month (Captain AI, custom branding, agent capacity management, priority email support) and Enterprise at $99/agent/month (adds SSO/SAML, SLA policies, phone support). Neither is required to run Chatwoot — the open-source edition is fully functional — but a 10-agent team that wants SSO and a support contract is looking at $99 × 10 × 12 = $11,880 a year. That's inside Conferbot's range, and unlike the DevOps-hours estimate, it's a real, published price for a real thing you'd be buying on purpose. It's also the same line item we ran into auditing Chatwoot's Captain AI against Intercom Fin: the free tier and the priced tier are genuinely different products, and a fair comparison has to pick one.

What this means if you're actually choosing

The question worth asking isn't "what does the internet say self-hosting costs" — it's a short checklist you can answer about your own team, not about the internet's:

• Do you already run other Docker services? If someone on your team already treats a VPS as one more thing to patch, Chatwoot is a marginal addition, not a new job.
• What's your real conversation volume? Chatwoot's own tiers (4 cores/10,000 conversations a day, 8 cores/20,000) tell you which box to rent — most teams evaluating a helpdesk are nowhere near either number.
• Do you need SSO or an SLA today, or in case someone asks? That's the one line item in this whole comparison with a real published price, and it's worth naming as a want, not folding into a vague "DevOps cost."
• Who answers a 3 a.m. outage? If the honest answer is "nobody yet," that's the real cost of self-hosting — not a dollar figure, a responsibility you don't currently have covered.

Answer those and the $8,000–$24,000 figure either evaporates (existing staff, Community Edition, no SSO requirement) or mostly confirms itself for a reason the comparison post never states. It's a familiar pattern: auditing what Intercom Fin actually bills for a "resolution" turned up the same shape, a plausible-sounding number that doesn't survive a check against the vendor's own documentation.

Where Replium fits — and where it honestly doesn't

We don't compete on this axis at all: Replium isn't self-hosted. It's a hosted, multi-tenant platform — sign up, get a subdomain, done — and there's no Docker Compose file we hand you to run yourself. If what you actually want is to own the deployment, keep the database in your own infrastructure, or run behind an air gap, we're not an option, and this audit doesn't change that; Chatwoot is a real answer to that need. Same honesty on data residency: we run a single production instance today with no per-region data isolation, so if the reason you're looking at self-hosting is keeping EU (or any specific region's) data on infrastructure you control, that's also a real reason to pick Chatwoot over us, not a gap we can argue you out of.

What we trade instead: no self-hosting bill of any size, because there's no self-hosting, and no DevOps hours, because that's on us. In exchange you get a tenant, not a server — you don't pick a Postgres version or SSH into anything, and there's no Enterprise tier to weigh against a support contract you might not need yet. If the trade you actually want is a smaller team with less infrastructure to think about, that's the trade we make explicitly, not one we back into with a comparison chart.