n8n for growing businesses

What n8n actually is

n8n is a tool for wiring systems together without programming everything yourself. You drag blocks onto a canvas: fetch something from system A, do something with it, write it into system B. Each block is a step, and the lines between them set the order.

What separates it from the better-known alternatives is that n8n is open source and can run on your own server. That sounds like a detail for engineers, but it has two consequences that matter in practice: your data never has to leave your environment, and you do not pay per action executed.

On top of that you can drop a piece of your own code at any point in a workflow. That seems unnecessary until you hit the one integration with no ready-made block, which happens regularly with regional business software.

Self-hosting or the cloud version

n8n comes in two forms: their own hosted version you pay for monthly, and the version you run on a server yourself. The choice turns on two things: how much you run and what passes through the workflows.

• Low volume and no sensitive data: start on the cloud
• Customer, order or staff data passing through: consider self-hosting
• Tens of thousands of actions a month: self-hosting gets cheaper quickly

When n8n beats Zapier or Make, and when it does not

Zapier is the easiest to start with and has the most ready-made connectors. For a handful of simple automations it is excellent, and if you already run it and it works, switching is rarely worth the effort.

It changes once volume grows. Zapier charges per task executed, and that bill climbs hard at thousands of actions a month. n8n does not meter per action, which makes it considerably cheaper at volume.

Make sits in between: cheaper than Zapier, visually more capable, but still a closed environment your data travels through. For companies with no objection to that, Make is often a fine choice.

Where n8n suits less well: when nobody in your organisation is technically inclined and you do not want to hire anyone for maintenance either. The tool gives you more freedom, and more freedom means more ways to build something nobody understands six months later.

What businesses actually build with it

The workflows that pay for themselves fastest are rarely spectacular. They are the connections someone currently makes by hand, every day.

• Pushing webshop orders through to accounting and the carrier
• Reading purchase invoices and queuing them for approval
• Synchronising stock levels between supplier, shop and marketplaces
• Enriching new leads from forms and dropping them into the CRM
• Assembling weekly reports from several sources
• Alerting when a value falls outside its expected range

AI steps inside a workflow

Since n8n gained blocks for language models, you can build steps that fixed rules cannot capture: classifying an incoming email, extracting an order from free text, summarising a long document.

The trap is leaning on that too heavily. A language model is not deterministic: the same input can produce different output. For classifying and summarising that is fine. For deciding an amount or picking a ledger account you want fixed rules, or at minimum a review step.

The rule of thumb I hold to: use AI to interpret messy input, use ordinary logic to make decisions with consequences.

Four mistakes I see most often

These come up almost every time with companies that started themselves and hit a wall after a year.

• No error handling: a workflow that stops silently, so weeks later it turns out orders were never processed
• Everything in one workflow: thirty steps in sequence that nobody can follow or test
• Keys hardcoded into the workflow instead of held in environment variables
• No backup of the workflows themselves, so one wrong click costs months of work

Doing it yourself or having it built

For a first simple integration, starting yourself is fine and I encourage it. The documentation is good and the community is active. Expect a few evenings before it clicks.

Bringing someone in becomes worthwhile once something business-critical runs through it, once several systems are involved, or once the person who built it is the only one who understands it. That last one is the real risk with in-house builds: not that it fails, but that it depends on one person.

Frequently asked questions about n8n