Zapier vs Make vs n8n: a decision tree for founders who don't code

The same workflow can cost $9, $24, or well north of $19.99 a month depending on which of these three tools runs it, and the difference has almost nothing to do with features. It comes down to what each platform decides to count. Zapier counts tasks, Make counts operations, and n8n counts executions, and those three words are the entire decision once you understand them.

Here is the one-sentence verdict on each, and the rest of this piece is the reasoning behind it.

Zapier: the default when your tools are niche or your team is non-technical; you pay a real premium per unit of work for the largest app catalog in automation (roughly 8,000 apps) and the gentlest editor in the category.

Make: the value pick for the middle of the market; full branching, iteration, and error-handling logic on a $9 tier, with a per-operation price that runs far below Zapier's per-task pricing at comparable tiers.

n8n: the graduation tool; it charges per workflow run rather than per step, self-hosts for free, and wins outright the moment your workflows get long, at the cost of asking you to read JSON when something breaks.

An automation platform, for the purposes of this article, is a tool that watches for an event in one app (a form submission, a new CRM deal, a webhook) and then performs actions in other apps without a human touching anything. All three do that job well. The question is what it costs and who fixes it when it stops.

The three tools side by side

	Zapier	Make	n8n
Pricing metric	Per task (each action a Zap performs)	Per operation (each module execution)	Per execution (one workflow run, any length)
Entry price	$19.99/mo (Professional, 750 tasks, annual billing)	$9/mo (Core, 10,000 operations, annual billing)	About $24/mo (Cloud Starter, 2,500 executions; roughly $20 on annual billing)
Free tier	100 tasks/month, unlimited two-step Zaps	1,000 operations/month, two active scenarios	Self-hosted Community edition, free and unmetered on your own server
Learning curve	Gentlest in the category; guided linear editor	Moderate; a visual canvas that rewards an afternoon with the docs	Steepest; items, expressions, and JSON in week one
App coverage	~8,000 apps	~2,000 apps	400+ native, plus an HTTP node for any API

Every figure above is the annual-billing rate as of August 2026, and every one of them moves if you pay monthly: Zapier's annual pricing runs roughly 20 to 33 percent below monthly, Make's saves 15 to 20 percent, n8n's about 17 percent. n8n also prices in euros, so the dollar figure floats a little with the exchange rate.

What the meters actually mean

This is the part most comparison articles skate past, and it is the part that decides your invoice.

A Zapier task is one action performed. Triggers are free, filters that stop a run are free, and Formatter steps are free, but every action that does something in an app consumes a task. A 7-step Zap that runs 100 times consumes 700 tasks. The number you need to model before choosing a plan is not how many Zaps you have; it is steps times volume. Zapier also auto-bills overage per task when you blow through your allowance, which protects your automations and surprises your budget in the same motion.

A Make operation is one module execution. Same shape as a task: a 6-module scenario that runs 500 times consumes 3,000 operations. The difference is the unit price. At entry, an operation on Make's Core plan costs about $0.0009, which runs far below Zapier's per-task pricing at comparable tiers. That arithmetic, not any feature, is the foundation of Make's value reputation.

An n8n execution is one run of a workflow, full stop. A 50-step workflow run bills as one execution where task-priced tools would bill 50. The longer your workflows, the bigger the gap. This is why n8n became the default engine for enrichment waterfalls of the kind we covered in our Clay-style pipeline playbook: those workflows are long by nature, and per-execution pricing makes their length free.

Notice what this means structurally. Zapier and Make both tax complexity: every step you add raises the bill. n8n taxes frequency: every extra run raises the bill, but the workflow can grow as sophisticated as you like at no marginal cost. Your automation style determines which tax is cheaper for you.

The same workflow, priced three ways

Take a workflow every founder eventually builds: a lead comes in from a form, gets enriched, gets scored, lands in the CRM, and pings a Slack channel. Call it six modules including the trigger, running 500 times a month, which is a healthy but unremarkable lead volume for a small business.

On Make, that is the textbook case from its own billing docs: a 6-module scenario that runs 500 times consumes 3,000 operations. That fits comfortably inside Core's 10,000 operations. Your bill is $9 a month on annual billing, and you have 7,000 operations of headroom before you think about the $16 Pro tier.

On Zapier, the trigger is free, so five actions times 500 runs is 2,500 tasks a month. The $19.99 Professional entry point includes 750 tasks, so this workflow alone consumes more than three times the entry allowance. You are either shopping the allowance slider well above the advertised entry price or watching auto-billed overage land on the invoice. Either way, this single workflow costs several times what Make charges for identical work, before you add a second one.

On n8n, 500 runs is 500 executions. The Cloud Starter plan includes 2,500 executions for about $24 a month, so this workflow uses a fifth of the allowance, and here is the important part: you could double the workflow's length, add a verification step, an AI qualification step, and a fallback branch, and the bill would not move. If you self-host the free Community edition, the license cost is zero and your ceiling is your server.

At this modest scale, Make wins on price and n8n wins on price-per-unit-of-ambition. Run the same math at 5,000 leads a month and the gap stops being a rounding error: 30,000 operations pushes you up Make's tiers, 25,000 tasks pushes Zapier into serious money, and 5,000 executions on n8n is one tier up from Starter. Long and frequent workflows are where n8n's meter becomes the only sane one.

So why does Zapier still own the category? Because the meter is not the whole decision.

The decision tree, in order

Work through these questions in sequence. Most founders get an answer by question two.

First: are your apps actually supported?

Before any pricing math, check the catalogs. Zapier connects roughly 8,000 apps; Make connects around 2,000; n8n has 400+ native integrations plus a generic HTTP node that can call any API if you are willing to read that API's documentation.

If your business runs on mainstream SaaS, all three will cover you and you move to the next question. If you depend on a long-tail tool, the local-industry CRM, the booking system nobody else uses, Zapier is routinely the only platform that connects it, and the decision tree ends here. This is not a small edge case. "Just Zapier it" became a verb precisely because the integration you need usually exists there and often exists nowhere else. Paying Zapier's premium for coverage you genuinely need is rational; paying it when all your tools sit in Make's catalog is buying breadth you will never use.

Second: who maintains this at 9 a.m. on a Tuesday?

Automations break. An API changes, a credential expires, a field gets renamed, and someone has to open the editor and figure out why leads stopped flowing.

If that someone is non-technical, Zapier's guided linear editor is worth real money. Setup walks you through connecting accounts, mapping fields with dropdowns, and testing each step against sample data; most first Zaps are live within ten minutes, and failure states are legible to a person who has never seen an API response. Make's canvas is approachable but its error-handler semantics take deliberate study. n8n confronts you with items, expressions, and JSON structure in week one, and a non-technical operator will stall exactly where Zapier would have carried them.

If the maintainer is technical, or even patiently semi-technical, the calculus flips. n8n's per-node execution logs show exactly what each step received and produced on the last run, which turns "why did row 412 fail" from archaeology into a click, and its code nodes mean there is no ceiling to hit. The honest way to say it: Zapier's premium is partly a salary substitute. If you have the skill in-house, stop paying for it twice.

Third: what shape are your workflows?

Now the meters matter. Sketch your three most valuable automations and count the steps.

Short and simple (two to four steps: form to CRM, payment to Slack, calendar to email): any platform handles these, and the free tiers may genuinely be enough. Zapier's 100 free tasks cover a handful of daily handoffs; Make's 1,000 free operations run a couple of real automations indefinitely. Pick on catalog and comfort, not cost.

Branching and batch-heavy (order lines that fan out to fulfillment and accounting, leads routed differently by segment): this is Make's home turf. Routers, iterators, aggregators, and error handlers are native drag-and-drop primitives on every tier including free, where Zapier's Paths and Looping exist but strain. You get genuinely programmatic control without code, at the category's lowest entry price.

Long pipelines (enrichment waterfalls, multi-provider fallbacks, AI qualification chains that run ten to fifty steps): n8n, and it is not close. Per-step metering, Make's included, punishes exactly this shape, while n8n bills the run. Add the LangChain-based agent nodes, which make n8n one of the fastest paths from AI idea to AI in production, and the graduation pattern writes itself: founders prototype the pipeline elsewhere, watch the meter, and rebuild it on n8n.

Fourth: does the data need to stay home?

A quieter question that occasionally overrides everything above. Zapier is cloud-only with US-based processing and no self-hosted or regional-residency option. Make is cloud-only but lets you choose an EU data center at signup on self-serve tiers, which satisfies most European compliance reviews without an enterprise contract. n8n self-hosts, full stop: every payload, credential, and log can live on your own server, which is the definitive answer for regulated teams and the strongest vendor-risk position in the category. If legal has opinions, this question moves to the front of the tree.

The traps, tool by tool

Each platform has one billing behavior that surprises people, and knowing them in advance is worth more than any feature comparison.

Zapier's trap is overage. When a Zap runs away, misconfigured trigger, unexpected volume spike, the platform auto-purchases tasks above your plan rather than stopping the automation. Your workflows keep working and your invoice explains later. Filter placement is your main lever, since filters that stop a run cost nothing; put them early.

Make's trap is the free tier's clock. The 15-minute minimum scheduling interval on the free plan means anything resembling real-time needs Core's 1-minute scheduling or webhooks, which are instant on every tier. And scenarios pause when the operation allowance runs out unless you configure top-ups, which is the opposite failure mode from Zapier: your budget is safe and your automation is dead.

n8n's trap is the word free. The Community edition has no license cost, but you supply the server, the upgrades, the backups, and the attention. A small VPS suffices for modest loads, and for a technical founder the trade is excellent. For everyone else, the roughly $24 hosted Starter plan is the real entry price, and it is still the cheapest way to run long workflows that exists.

Where each verdict lands

Run the tree honestly and the three tools stop competing, because they serve three different founders. The founder with niche tools or no technical bench pays Zapier's premium knowingly and gets the fastest time-to-first-automation in the category from a famously durable company. The cost-conscious founder with mainstream apps and moderately complex flows takes Make's $9 tier and gets 90 percent of what either neighbor offers at clearly the lowest bill. The technical founder building long pipelines takes n8n and gets economics nobody else can match plus a free exit ramp to their own server.

One migration note, because it changes how you should decide: none of the three imports from the others. Leaving any of them means rebuilding workflows by hand, so the switching cost grows with every automation you add. Teams usually report the rebuild is faster than the original build, since the logic is already designed, but the asymmetry is real. It costs nothing to enter Zapier and something to leave it. Decide with the tree now rather than after twenty Zaps are load-bearing.

If you want the deeper per-tool teardowns, all three dossiers live in our GTM engineering category, alongside Serper, the search API that tends to show up inside the pipelines these tools run.

The short version

Check the app catalog first; if only Zapier connects your tools, you are done. Otherwise, match the meter to your workflows: Make's $9 per-operation pricing wins for short and branching automations maintained by non-engineers, and n8n's per-execution pricing wins decisively once workflows get long or you can self-host. Model steps times volume on all three before you commit, because the platforms price the same work as differently as $9 and a multiple of $19.99, and rebuilding later is manual.