Most people start comparing automation tools on a bad day. A lead form stopped reaching the CRM sometime last week and nobody noticed. Or the monthly bill jumped because one Zap looped over a spreadsheet with four thousand rows. Or the person who built every workflow left, and the rest of the team is afraid to open the editor.
On a day like that, the question "Zapier vs Make vs n8n" is really two questions. Which tool will cost less for the same work, and which one will break less often, or at least tell you when it does? This post answers both. It covers how each of the three bills, what each does when a run fails, the three head-to-head matchups people search for, and how to choose. The last part covers what to do when the problem is that nobody wants to maintain automations at all, which is the case for an AI employee. We make one, called Sanaf, so read that section knowing we have an interest. Everything before it holds whichever tool you pick.
The Zapier and n8n details below were checked against their own pricing pages and help docs on 3 November 2026. Make's pricing page wouldn't load for our checker, so its billing rules come from the Make Help Center and we haven't printed a Make plan price. Prices on all three change often, so treat any number here as the starting point for your own check.
The short answer
Zapier, Make and n8n all connect your apps and run a workflow when something happens. Zapier has the most connectors (over 9,000) and the gentlest learning curve, and it bills per task, meaning every successful action step. Make has the most visual editor, bills in credits (one credit per operation for ordinary apps), and puts unlimited users on every plan. n8n is the most technical, bills per workflow execution however many steps that run has, and has a free self-hosted Community Edition. Choose Zapier if the app you need is only supported there or nobody on the team is technical. Choose Make if your workflows branch a lot and you want to see them. Choose n8n if you have someone technical and want either control or a lower bill at volume.
Zapier vs Make vs n8n at a glance
| Zapier | Make | n8n | |
|---|---|---|---|
| What you build | Zaps: a trigger and a list of steps | Scenarios on a visual canvas | Workflows of nodes on a canvas |
| What you pay for | Tasks (successful action steps) | Credits (1 per operation for non-AI apps) | Executions (one full run, any number of steps) |
| Free option | Free plan, 100 tasks a month, two-step Zaps | Free plan with a credit cap | Community Edition, self-hosted |
| Hosting | Zapier's cloud | Make's cloud | n8n's cloud, or your own server |
| App catalog | 9,000+ apps | Large module library | Large node library, plus HTTP and code for anything else |
| Who it suits | Non-technical teams, long-tail apps | Visual thinkers, complex branching | Technical teams, high volume, data control |
| When a run fails | Errored run in Zap history, Autoreplay on every plan | Incomplete executions, if you turn them on | Execution log, plus an error workflow you build |
How each one bills, and why the same workflow costs different amounts
The pricing models are the biggest practical difference between the three, and they're easy to misread. Take one ordinary workflow as the example: a new order arrives, you look up the customer, update a spreadsheet row, and send a Slack message.
Zapier pricing: tasks
Zapier counts a task as "any successful action that runs in Zapier" (Zapier Help Center). Triggers don't count. Filter and Paths steps don't count, and neither do steps that error or get skipped. So the order workflow above uses three tasks per order: the search, the row update and the message. A hundred orders a month is 300 tasks.
On the pricing page the Free plan includes 100 tasks a month and two-step Zaps, and checks for new data every 15 minutes. Professional starts from $19.99 a month with multi-step Zaps, webhooks and a two-minute check. Team starts from $69 a month with 25 users and shared connections, and Enterprise is quoted by sales. The price climbs with the task tier you pick, and paying yearly saves 33% on the page we checked. Answering "how much does Zapier cost" for your own business means counting action steps, then multiplying by how often the trigger fires.
Two details catch people out. AI steps and Code by Zapier steps carry their own task rates, and from 1 December 2026, Zapier says each successful run of a Next Gen Zap uses at least one task even when it only filters or formats. If your Zaps are mostly filters with one action at the end, read the current help page before you estimate.
Make pricing: credits
Make used to bill in operations and now bills in credits. Its help center says the change of name didn't change existing plans, and that for non-AI apps "1 operation equals 1 credit" (Make Help Center). An operation is roughly one module doing its work on one bundle of data, so the order workflow above uses about four credits per order, because the trigger module counts too. Every plan includes unlimited users, which is fair to a team where several people build.
The detail worth knowing is the overage. If you run out of credits mid-month, extra credits cost 25% more than the ones in your plan, whether you buy them by hand or let Make buy them automatically (Make Help Center). Make itself recommends moving to a bigger plan if you regularly buy extras. Check make.com/en/pricing for the current plan prices and credit tiers.
n8n pricing: executions
n8n bills differently. Its pricing page says pricing "is based on monthly workflow executions, regardless of complexity" (n8n pricing). One execution is one full run of a workflow, so the order workflow is one execution per order whether it has four steps or forty. Every plan includes unlimited users, unlimited workflows and every integration.
When we checked, the page showed prices in euros on annual billing: Starter at €20 a month for 2,500 executions, Pro at €50 a month for 10,000, and Business at €667 a month for 40,000, with Business self-hosted only. Enterprise is quoted by sales. There's also a startup discount of 50% off Business for companies under 20 people that qualify. Your currency and the monthly price may differ.
The execution model is why n8n gets cheap at volume. A workflow with ten action steps that runs a thousand times a month is 10,000 tasks on Zapier, around 11,000 credits on Make, and 1,000 executions on n8n. The flip side is a limit worth reading: Starter allows five concurrent executions and a maximum run time of five minutes.
Is n8n free?
Partly. The Community Edition is free and open source, and you host it yourself (n8n on GitHub). n8n self hosted means you pay for the server, keep it updated, back it up, and fix it when it falls over at 2am. For a technical team that's a fair trade, and it keeps your data on infrastructure you control. For a team with nobody to run a server, the free version costs more in time than the Starter plan costs in money. The hosted n8n cloud plans come with a free trial and no card.
Why automations keep breaking
Comparing pricing assumes the workflows work. The more common complaint is that they stop. In our experience the causes are the same on all three tools:
- A connection expires. Someone changes a password, an app revokes a token, or the person whose login the workflow used leaves the company. Every run after that fails.
- The data changes shape. A form gets a new field, a spreadsheet column is renamed, or a customer types a phone number with letters in it. The step that expected the old shape errors, or worse, runs and writes the wrong thing.
- An app changes its API. The connector gets updated and a field you mapped is gone or renamed.
- You hit a limit. The plan runs out of tasks or credits, a run takes longer than the plan allows, or too many items arrive at once and flood protection holds them.
- The case was never drawn. The workflow handles the three situations its builder thought of. The fourth one, a refund on a split order or a lead with two email addresses, goes down whatever branch is nearest.
The first four produce errors, which at least show up somewhere. The fifth produces runs that succeed and do the wrong thing, and no tool's error handling will catch those, because nothing failed.
What each tool does when a run fails
This is where the three differ most in day-to-day life, and it matters more than the connector count once you have twenty workflows running.
Zapier
A failed run shows as Errored in your Zap history, and Zapier can explain the error with an AI troubleshooting tab (Zapier Help Center). If a Zap errors repeatedly, Zapier turns it off automatically. Runs that stop because an app disconnected or you hit your task limit go On hold instead, and Zapier keeps the trigger data so you can replay them once you've fixed the cause.
Autoreplay is available on every plan, including Free. Turn it on and Zapier retries a failed step up to five times, with the gap growing each time, so the last attempt lands about 10 hours and 35 minutes after the first error (Zapier Help Center). It won't email you about the error until the last attempt fails. Replays have to happen within 60 days of the trigger, and you need enough tasks left to cover them. Zapier also supports custom error handling, an alternative path that runs when a step fails.
Make
Make's safety net is called incomplete executions. When a scenario hits an error with them switched on, Make stores the unfinished run so you can retry it automatically or resolve it by hand. The catch is in the first line of the docs: incomplete executions are disabled by default, and you turn them on per scenario in its settings (Make Help Center).
There are gaps even when they're on. An error on the first module doesn't create an incomplete execution unless you add a Retry error handler to it. And if your storage for incomplete executions fills up, Make either disables the scenario or, if you allowed data loss, keeps running and throws the failed runs away (Make Help Center). Make also has proper error handler modules (retry, ignore, rollback and others) you can attach to any module, which is powerful once you learn them.
n8n
Every run appears in the executions list, and you can load data from a past run into the editor to debug it. For alerts, you build an error workflow: a separate workflow that starts with the Error Trigger node and runs whenever another workflow fails, which you might use to post to Slack or send an email (n8n docs). One error workflow can cover many workflows. Automatic retries are included on the cloud plans, and steps can be set to retry on failure.
That's the n8n pattern in general. You can build exactly the error handling you want, and nobody builds it for you.
Zapier vs Make
Make vs Zapier is the comparison most people run first, because they're the two easiest to start with.
Zapier is quicker to learn. A Zap reads top to bottom like a list, the app catalog is the largest of the three, and most people have a working Zap in their first ten minutes. Make asks a bit more up front, because a scenario is a canvas of modules with routers and iterators, but that canvas is the reason people switch. A workflow with five branches is easier to read in Make than in a Zap with nested Paths.
On cost, it depends on shape. Zapier doesn't charge for triggers or filters, which helps workflows that filter out most items. Make charges for the trigger module but includes unlimited users, which helps a team where four people build. Price your three busiest workflows on both before deciding.
Pick Zapier if you need a connector only Zapier has, or if the people building aren't technical. Pick Make if your workflows branch, loop or transform data a lot, and you want to see every step.
n8n vs Zapier
n8n vs Zapier is the most searched of the three matchups, and it's really a question about who's on your team.
Zapier is the one you can hand to anyone. n8n is the one a technical person enjoys: it has code steps in JavaScript and Python, custom HTTP requests to any API, and a self-hosted option that keeps data on your own servers. The trade is that n8n assumes someone will own it. If nobody does, the workflows stay exactly as they were built while the apps around them change.
On cost, n8n usually wins for long workflows that run often, because a forty-step run is one execution. Zapier can win for small, simple Zaps where the free or entry plan covers you. Zapier's catalog is larger, but n8n's HTTP node lets a developer reach almost any API that lacks a ready-made node, which narrows the gap in practice for teams who can use it.
Pick Zapier if no one on the team writes code. Pick n8n if someone does, and either your volume is high or your data has to stay on infrastructure you control.
n8n vs Make
Make vs n8n is the closest matchup, because both are visual canvases aimed at people willing to learn a tool properly.
Make is easier to start and fully hosted, with a long list of ready modules. n8n is more flexible underneath: code in any step, self-hosting, and AI agent nodes with human approval for tool calls. Billing is the cleanest split. Make counts operations, so a long scenario costs more per run. n8n counts executions, so it doesn't.
Pick Make if you want a hosted tool and nobody wants to touch code. Pick n8n if you want to self-host or your scenarios are long and frequent enough that per-operation billing hurts.
How to choose between Zapier, Make and n8n
Run through these in order and stop at the first one that decides it:
- Is an app you depend on only supported in one tool? Then use that tool. Check each catalog for your exact apps and the exact triggers and actions you need before anything else.
- Does the data have to stay on your own servers? Then n8n self hosted is the only one of the three that does that.
- Is anyone on the team technical and willing to own the workflows? If not, Zapier is the safest choice, and Make is fine if the workflows are simple.
- How much volume do you run? Count runs and steps for your three busiest workflows and price them on all three. Long, frequent workflows favor n8n. Short, filtered ones can favor Zapier.
- Who will notice when it breaks? Whatever you choose, turn on the safety net on day one: Autoreplay and error alerts in Zapier, incomplete executions and error handlers in Make, an error workflow in n8n.
Our Zapier alternatives page covers the same ground from the other side, for people already on Zapier and deciding whether to leave.
When the automation itself is the problem
There's a case none of the three tools fixes. Sometimes moving from Zapier to Make, or from Make to n8n, just means rebuilding the same fragile workflows somewhere cheaper. The work keeps breaking because it isn't really trigger-shaped. It needs someone to read what arrived and decide what to do with it: a customer email that's half complaint and half order change, a lead whose details don't match the form, an invoice that's almost right.
That's the line between automation and an AI agent. An automation follows a path drawn in advance, which is exactly what you want for moving a form submission into a spreadsheet. An AI agent is given a goal and works out the steps, which suits work that varies case to case. Our post on AI agents vs chatbots, assistants and automation goes through that distinction in more detail.
Most businesses end up wanting both. The plumbing that never changes stays in Zapier, Make or n8n, where it's cheap and predictable. The work that needs reading and judgment goes to something that can read and judge, with a person approving anything that matters.
Where Sanaf fits
Sanaf is an AI employee. It works in your team's Slack or Microsoft Teams, and inside the apps you connect. You don't draw a workflow. You describe a job in plain words, such as "when an order email comes in, check it against the order in our store, fix anything that doesn't match and tell me what you changed", and set what it may do without asking.
The difference from the three tools above shows up when something unexpected arrives. A Zap or scenario either errors or follows the nearest branch. Sanaf reads the message, works out what's going on, and asks in the thread when it isn't sure. It asks before anything it can't undo, like sending money or deleting a record, and keeps a work log your team can read, written like a colleague reporting back rather than an execution history. Nobody has to maintain error branches, because there aren't any to maintain.
It isn't the right tool for everything. If you need a high-volume, fixed pipeline that runs the same way ten thousand times a day, a workflow tool is cheaper and more predictable, and we'd say so. Our comparisons with Zapier Agents, Make and n8n lay out where each one wins. Check the integrations directory for the apps you need before you start.
On price, you start with $120 of free credit and no card, and that free credit lasts 30 days. After that you buy credits from $20 when you need them, and credit you buy never expires. You pay for the work, not per seat. The pricing page has the details, and hiring an AI employee shows what the first day looks like.
A good first test: pick the workflow that broke most recently, describe what it was supposed to do in two sentences, and give that job to Sanaf for a week alongside the old one.
Questions people ask about Zapier, Make and n8n
Which is better, Zapier, Make or n8n?
None is better for everyone. Zapier is best for non-technical teams and long-tail apps, with over 9,000 connectors. Make is best for complex, branching workflows you want to see on a canvas. n8n is best for technical teams who want to self-host or run long workflows at high volume, because it bills per execution rather than per step.
Is n8n cheaper than Zapier?
Usually, for long workflows that run often, because n8n charges one execution per run however many steps it has, while Zapier charges a task for every successful action step. For a few short Zaps, Zapier's free or entry plan can cost the same or less. Price your own busiest workflows on both.
Is Make cheaper than Zapier?
It depends on the workflow. Make charges a credit per operation, including the trigger module, but includes unlimited users on every plan. Zapier doesn't charge for triggers, filters or failed steps. A team with several builders often pays less on Make, and a workflow that filters out most items can cost less on Zapier.
Is n8n free?
The self-hosted Community Edition is free and open source, but you pay for the server and the time to run it. The hosted n8n cloud plans are paid, priced by monthly executions, with a free trial that needs no card.
How much does Zapier cost?
Zapier has a Free plan with 100 tasks a month and two-step Zaps. Professional starts from $19.99 a month and Team from $69 a month, and the price rises with the number of tasks you choose. Enterprise is priced by sales. Check zapier.com/pricing for current figures.
What happens when a Zap fails?
The run shows as Errored in your Zap history. If you turn on Autoreplay, Zapier retries the failed step up to five times over about ten and a half hours. If a Zap errors repeatedly, Zapier turns it off automatically, and runs held by a disconnected app or a task limit can be replayed after you fix the cause.
Can I use Zapier, Make and n8n together?
Yes, and some teams do, keeping a connector that only one tool has in that tool. Each extra platform is one more place to check when something breaks, so most teams settle on one for workflows and add an AI employee for the work that needs judgment.
What is the best Zapier alternative?
It depends on why you're leaving. For a lower bill, Make or n8n. For a nicer visual editor, Make. For self-hosting, n8n. If maintaining workflows is the problem, an AI employee like Sanaf takes a job described in words instead. The Zapier alternatives page compares four options.