Power Automate vs Make vs n8n vs custom code: which one should a small business actually use?
An honest comparison of the main automation tools from someone who uses all four, with a simple rule for picking one based on your stack and how much failure you can tolerate.
Every automation vendor says they can connect anything to anything. Technically most of them can. The question is which one will still be working quietly at 6am on a Monday in eight months.
The short version
| If you are… | Start with |
|---|---|
| A Microsoft 365 shop (Outlook, SharePoint, Teams, Dynamics) | Power Automate |
| On a mixed stack of SaaS tools and want to move fast | Make |
| Technical, cost-sensitive, or handling data you cannot send to a US cloud | n8n (self-hosted) |
| Running high volume, legacy systems, or anything where a silent failure costs real money | Custom code |
Power Automate
Included in most Microsoft 365 licences, which is its biggest advantage: there is nothing new to buy and IT already trusts it. Excellent for anything that starts in an inbox or ends in SharePoint. Weak on error handling by default, and the pricing gets confusing the moment you need premium connectors. Build in retries and a failure notification from day one.
Make
The best visual builder of the group and a very large connector library. Pricing is by operation, which is predictable at low volume and surprising at high volume. Good for prototyping a workflow in a day. Watch out for scenarios that grow into forty modules nobody wants to touch.
n8n
Open source, self-hostable, and happy to run custom JavaScript inside a node. Data stays on your infrastructure, which matters for regulated industries and European clients. The trade-off is that someone has to look after the server.
Custom code
A small Node or Python service with a database, a queue and proper logging. More expensive up front, dramatically cheaper per run at volume, and the only option that gives you full control over what happens when something fails. This is where inventory sync for a business doing thousands of orders a month usually ends up.
The rule I use
Pick the tool that your team already pays for and that IT already allows, unless the cost of a silent failure is high. When it is, write code. A broken no-code flow that nobody noticed for three weeks is far more expensive than the developer time it would have taken to build it properly.
If you are not sure which bucket you are in, that is the first thing we work out on a Workflow Audit.