Your Shopify store gets an order. Someone has to copy the customer into HubSpot, ping Slack, update a Google Sheet, and maybe trigger an invoice in QuickBooks. You could pay Zapier by the task — or you could run n8n on a server you control.
That's the pitch, anyway. n8n is a visual workflow tool: drag nodes onto a canvas, connect them, wire up credentials once, and let the thing run. It's been around long enough to feel mature (190k+ GitHub stars doesn't happen by accident) and new enough that the AI nodes aren't just marketing stickers on a 2019 product.
What n8n actually does
Think Zapier or Make, but you install it. You build workflows — chains of steps triggered by a webhook, a schedule, or an event in another app. Each step might read a database, call an API, transform JSON, send an email, or run a bit of JavaScript when the built-in nodes aren't enough.
The integration list is long: Shopify, Stripe, Slack, Notion, PostgreSQL, Gmail, GitHub, OpenAI, and hundreds more. Agencies love it for gluing client stacks together without writing a custom integration for every job. Ops teams use it for alerts, data sync, and the boring stuff that used to live in a cron job nobody documented.
Why self-host instead of n8n Cloud?
n8n Cloud is fine for prototyping. Production is a different conversation.
Self-hosting means your credentials live on your infrastructure — not in a multi-tenant SaaS where your Stripe keys and CRM tokens share a platform with everyone else. For Canadian businesses thinking about PIPEDA, that's often the deciding factor: your workflow logic, your API keys, and the customer data flowing through those pipes stay in a Canadian data centre if you put the server there.
Cost is the other big one. Cloud plans charge per execution. A busy store or a chatty CRM sync can burn through limits fast. On your own VPS, you're paying for CPU and RAM — not per webhook fired.
And you can reach internal systems. n8n on your network can talk to a private PostgreSQL instance, an internal REST API, or a VPN-connected tool that no public SaaS automation platform will ever touch.
What running it takes
n8n runs happily in Docker — that's how most people deploy it. For a small team getting started, a modest VPS with 4 GB RAM is a reasonable starting point. Add PostgreSQL instead of SQLite once you're serious (SQLite is fine for tinkering; you'll want Postgres when workflows multiply).
When volume picks up, enable queue mode with Redis so executions don't pile up and die during a spike. HTTPS webhooks are non-negotiable in production — most integrations expect a proper TLS endpoint. Back up the database and your encryption key; lose the key and your stored credentials are gone for good.
License note: n8n uses a fair-code model — source-available, free to self-host for your own use, with restrictions if you're reselling n8n as a service. For internal automation on your own server, that's typically a non-issue. Read their license if you're an MSP planning to host it for clients.
Who it's for (and who should skip it)
Good fit: marketing and ops teams tired of per-task SaaS bills, agencies connecting client tools, SaaS companies wiring internal alerts, anyone who needs Canadian data residency for automation logic.
Maybe skip it: if you need one simple Zap and will never think about it again — Cloud is easier. If nobody on your team will maintain the server, unpaid SaaS might beat an unmaintained self-hosted instance.
Hosting it in Canada
We host n8n for clients who want queue mode, Postgres, backups, and someone to answer when a workflow breaks at 2 a.m. — on Canadian VPS infrastructure, documented and monitored. If you're already running Docker Compose stacks, n8n drops in cleanly alongside your other services.
See our n8n hosting page for what's included, or tell us about your workflows and we'll size an environment that matches how many executions you're actually running — not a generic tier chart.
