Choosing a platform
Make vs Zapier vs n8n: which is cheapest for a small business?
The three differ far less in features than the comparison pages suggest, and far more in how they charge you. That single difference is what decides your bill — and it is the thing almost nobody explains before you have already built twenty automations inside one of them.
The short answer
If you want to be running today and your volumes are modest, use Zapier. If you expect to build a dozen automations and keep them running, use Make. If you are technically confident, handle sensitive data, or run real volume, use n8n.
If you genuinely cannot decide, use Make. It is the middle answer on every axis, which means being wrong about it costs the least.
That is the whole recommendation. The rest of this explains why, because the reasoning matters more than the answer — vendors change their pricing, and you will need to make this judgement again.
What these things actually do
Your website form does not know your spreadsheet exists. Your spreadsheet has never heard of your AI tool. Your email does not know either of them. These products were built by different companies and have no reason to talk to one another.
An automation platform is the wiring in between. It watches one service, and when something happens there, it does things in the others: reads a record, calls an AI tool, chooses a path, writes an email, logs a line.
Before these existed, connecting two applications meant employing somebody to write code. That is the entire purpose of the category — it lets you wire services together without programming.
The row that actually decides your bill
Features converge. Every one of these will do what a small business needs. Where they differ genuinely is who they are built for and — the part that costs you money — what counts as one billable unit.
| Zapier | Make | n8n | |
|---|---|---|---|
| Feels like | Filling in a sentence | Drawing a diagram | Building a machine |
| Learning curve | Gentlest | Moderate | Steepest |
| Charges by | Task — each action step that runs | Operation — each module that runs | Execution — each whole run |
| Cost at volume | Highest | Middle | Lowest |
| Self-host | No | No | Yes — free software, you pay for a server |
| Suits | Getting running today | Branching, seeing the shape | Volume, control, technical comfort |
Read the third row twice. A six-step automation running once costs roughly six tasks on Zapier, six operations on Make, and one execution on n8n.
That difference is invisible while you are learning and enormous once something runs a few hundred times a day. It is the single most common reason businesses migrate away from a platform after a year — not because they disliked it, but because the arithmetic changed underneath them.
You will notice there are no prices here. Prices change, and an article that quotes them is wrong within a year and misleading before that. Check each vendor's current pricing page yourself — what does not change is how each one counts.
Zapier — the easiest place to start
The oldest and largest, and it shows in two ways: it connects to almost anything, and it is the least intimidating to use. You choose a trigger, choose an action, and it reads almost like a sentence — when this happens, do that.
Good for your first automation, for unusual applications nobody else supports, and for anyone who wants to be running by this afternoon rather than understanding a canvas first.
Where it hurts is price. It charges per task, so a multi-step automation consumes several units every time it runs, and the plans climb steeply. Businesses that grow into real volume commonly move away for exactly this reason.
Make — the middle ground, and the one you can see
Make shows your automation as boxes joined by lines on a canvas. You watch the data travel through it step by step, and when something fails you can see precisely where.
That visual model is worth more than it sounds. Most people who struggle with automation are not struggling with the tool — they are struggling to picture what is happening. A canvas removes that problem in a way a list of steps does not.
Good for anything with branches, anything that loops over a list, and for understanding what your automation is actually doing. It costs an hour or two more learning than Zapier, and the canvas is genuinely confusing for the first twenty minutes. Then it is not.
n8n — the technical one, cheapest at scale
n8n is open-source. You can use their hosted service, or run the software on your own server and pay only for the server — which at volume is dramatically cheaper than anything else here.
It also charges differently, and this is the part worth understanding. An execution is one whole run, however many steps it contains. A twenty-step automation costs the same as a two-step one. For complicated work, that changes the arithmetic completely.
Where it hurts is that it assumes more of you. Self-hosting means you are responsible for the server, the updates and the backups. If that sentence sounded like work rather than an opportunity, this is not your starting point.
Two things to check before you commit
- Does it connect to your systems? Not the famous ones — your booking system, your accounting package, your particular shop. Look each one up specifically. This decides it more often than anything else on this page.
- Build one real automation on the free tier first. All three have one. Forty minutes with a real tool tells you more than any comparison, including this one.
What switching later actually costs
Less than you fear. What you do not lose is the understanding: triggers, data lookups, branches, actions and approval gates exist in all three, and the way you think about an automation transfers completely.
What you lose is the building. Ten automations is perhaps a day of rebuilding, from notes you already have. Unpleasant, not disastrous.
So do not agonise over this decision. The cost of choosing imperfectly is one day, some time from now. The cost of not choosing is every week you spend doing the work by hand in the meantime.
The decision nobody mentions
Which platform you pick is the second most important choice you will make. The first is deciding, for each job, what the automation is allowed to do without you.
There are three sensible answers. It drafts and you send. It acts, but only once you have said yes. Or it runs unattended and you find out afterwards. That choice determines what happens on the day something goes wrong — and something will, on every platform equally.
An automation that drafts a reply for you to check is safe on any of these three. One that emails customers unattended is a risk on all three. The platform does not change that; only your decision does.