n8n is a workflow tool in the same family as Zapier and Make, but it lets you go further: branching logic, your own code where you need it and the choice to run the whole thing on your own server. That flexibility is why people hire an n8n consultant. It is also why a badly built n8n setup is harder to rescue than a badly built zap. If you are searching for an n8n automation expert or an n8n workflow developer, these are the questions that separate the good ones from the rest.
Six questions to ask before you hire
- What happens when a workflow fails? The answer should be specific. n8n lets you set an error workflow for each one, which must start with an Error Trigger node and receives details of the failed execution. If a consultant doesn’t mention this, keep looking.
- Where will it run? Cloud or your own server, with the work that choice creates written down.
- Who can see the credentials? API keys and logins should be held inside n8n and shared deliberately, not pasted into workflow steps.
- How will I know what is running? You want a list: each workflow, what starts it, what it touches and who owns it.
- Can someone else pick this up? Naming, notes and a short handover call matter more than clever design.
- What are you not going to automate? Good consultants have a short list of things that need a person.
Cloud or your own server
n8n’s hosting documentation describes self-hosting on your own infrastructure, on-premises or in a private cloud, with Docker Compose among the install routes. Without a licence key, a self-hosted install runs as the free Community edition. Cloud is the hosted alternative, and the pricing page says Cloud plans are measured by monthly workflow executions.
A typical first project
Say a form on your site feeds a spreadsheet and somebody copies each row into the CRM by hand. We build a workflow that catches the form, looks the person up so you don’t create duplicates, writes the record and tells your team in chat. Then the part most people skip: we deliberately break it, for example with a wrong field name, and watch the alert arrive. n8n’s Stop and Error node can force a failure on purpose, which makes that test easy. Only then do we let it loose on real submissions, and we watch the first runs in the executions list to confirm each one did exactly what you expected, including the awkward ones like a missing phone number or a name in capitals.
What a consultant should hand back
| Item | Why it matters |
|---|---|
| Workflow list | So nobody asks “what does this one do?” in a year |
| Error routing | Failures reach a person, not a log nobody reads |
| Credential notes | Whose account each connection uses, and what happens when they leave |
| Run check | Where to look in the Executions view when something seems off |
| Switch-off steps | How to pause everything safely |
When n8n is the wrong tool
- You need one simple link between two apps, where a connector with a free tier is faster.
- Nobody on your side will ever open the editor, so the first change request becomes a support ticket.
- The process is so critical or so large that it deserves proper software, not a chain of nodes.
Rather build it yourself?
n8n is friendly enough to start on your own. Our guide to setting up n8n for your team walks through Cloud, roles and a first scheduled workflow, and our Zapier page covers the simpler alternative. Come to us when the logic gets tangled or the stakes rise. For the sibling tool in the Microsoft world, see our Power Automate consultant page.
A workflow that fails loudly is an asset. One that fails quietly is a liability you haven’t found yet.