
The cheapest of three quotes for the same automation is regularly the one that produces the most expensive month. One reads $19.99 for 750 tasks. Another reads $9 for 10,000 credits. A third reads 50 EUR for 2,500 executions, and you are meant to choose by Thursday. Nothing about those numbers is deceptive. The cause is a unit conversion error, and it survives because most pages listing Zapier alternatives print the sticker and skip the meter underneath it. You have probably already picked the one that looked cheapest. Which number would you have to check before you trusted that pick?
Zapier's own reporting, via a statistics compilation, puts 3.1 billion tasks a month through the platform, and a task is one successful action. That unit is not printed at the point of purchase, so a workflow costing nine units here costs one there, and the gap compounds in favour of whoever counts runs.
Why "Cheaper Than Zapier" Is Usually a Unit Conversion Error
Zapier counts a task: one successful action inside a Zap. Triggers and polling cost nothing, and the built-in Filter, Formatter, Paths, Tables and Forms steps are not charged. Make counts a credit, one standard module run, though a module reaching for an AI provider can cost more. n8n counts an execution, the whole workflow from start to finish regardless of how many nodes sit inside it. Activepieces counts a credit per flow run. IFTTT counts a credit per Applet run, MCP call or AI feature use. Power Automate counts actions against a daily ceiling that resets each morning.
Six meters, and only three count the same event.
Tool | Unit | One unit buys | Charged when | Not charged |
|---|---|---|---|---|
Zapier | task | one successful action inside a Zap | on success only | triggers, polling, Filter, Formatter, Paths, Tables, Forms steps |
Make | credit | one standard module run; AI-provider modules can cost more than one | when the module executes | Router and error-handler modules |
n8n | execution | the whole workflow, start to finish, any number of steps | when a workflow runs to completion | every node inside the run |
Power Automate | action, against a daily ceiling | one connector action | against a limit that resets daily | nothing bankable; unused capacity does not carry |
Activepieces | credit | one flow run, however many steps | per flow run | the steps inside the run |
IFTTT | credit | one Applet run, MCP call, or AI feature use | per use of any of those | nothing extra; all features sit on every tier |
What Zapier's own ladder does to a bill is worked through in Zapier pricing, billing toggle included.
Now the number that decides this. Run a ten-step workflow once and Zapier bills nine tasks, because the trigger is free and each action is not. Make bills about nine credits. n8n bills one execution. Activepieces and IFTTT bill one credit each. Power Automate bills nine actions against the day's ceiling.
Run that workflow 5,000 times in a month and it is 45,000 Zapier tasks against 5,000 n8n executions. Zapier's 5,000-task tier does not see the workload at all, while n8n's 10,000-execution tier absorbs it with half the allowance spare, for 50 EUR a month on the plan card. A two-step Zap collapses the gap to one unit against one, which is why anyone who only ever built two-step Zaps is right that the difference is small.
The Cost Table
Every figure was read on 2026-11-20. The workload: 5,000 units a month, small enough for every tool here to serve and large enough for tier structure to show.
Tool | Unit | Smallest tier covering 5,000 | Price at that tier | Cost per unit used | Self-host |
|---|---|---|---|---|---|
Zapier | task | Professional 5,000 tasks | $89.00/mo annual, $133.50 monthly (secondary) | $0.0178 | No |
Make | credit | Core 10,000 credits | $9.00/mo annual | $0.0018 | No |
n8n | execution | Pro 10,000 executions | 50 EUR/mo annual | 0.0100 EUR | Yes |
Power Automate | action | Premium, one user | $15.00/user/mo (secondary) | $15.00 for the seat; 0.4% of the daily allowance | No |
Activepieces | credit | Plus 10,000 credits | $20.00 flat | $0.0040 | Yes |
Pipedream | credit | Startup 10,000 credits | $99.00/mo annual, $150 monthly | $0.0198 | No |
IFTTT | credit | Starter 5,000 credits | $2.09/mo annual | $0.00042 | No |
Where a tier overshoots 5,000, the per-unit figure divides the price by 5,000 used, not by the allowance bought. Make's Core tier charges $9 for 10,000 credits, so a 5,000-credit month is $0.0018 a credit used with half the allowance untouched, or $0.0009 per credit bought.
Three rows carry a provenance note. Zapier prints its 5,000-task tier only once the slider is operated, and two independent pages dated October 2026 agree on $89 annual and $133.50 monthly. Microsoft's page rendered without readable amounts, so $15 per user a month is secondary-sourced. Make's own pricing page returned HTTP 403 behind a bot wall, so its ladder comes from Make's comparison page.
Read across and the spread is 47 times, from IFTTT at $2.09 to Pipedream at $99, and that comparison is the error this article exists to name. IFTTT's credit buys one consumer Applet run with no branching and no code. Zapier's task buys one action inside a multi-step workflow with 9,000 connectors behind it.
What Each Zapier Alternative Is Actually For
Six tools, one question each: what this costs at your workload, and where it stops. Read each section for the constraint at the end, not the price at the top.
Make: 10,000 Credits for $9, and Half of Them Go Unused

Make is the cheapest place to put a complicated workflow you do not want to host. Core is $9 a month billed annually for 10,000 credits, Pro is $16, Teams is $29, and every paid tier starts at that same 10,000. The ladder comes from Make's own comparison page, because its pricing page returned HTTP 403. The unit also moved: as of 27 August credits are the billing unit, per Make's pricing FAQ, replacing operations. Most modules consume one, AI-provider modules several, Router and error handlers none.
Where it stops. Credits expire at the end of the term, so the allowance is a deadline, not a balance. Make is cloud only, which its own comparison page calls "cloud-first at all subscription levels".
n8n: One Execution Buys the Whole Workflow

n8n is the only tool here whose meter ignores your workflow size. One execution is one run, start to finish, so a 40-node workflow and a two-node one cost the same. Starter is 20 EUR a month billed annually for 2,500 executions; Pro is 50 EUR for 10,000 on the plan card, while the same page's limits table prints 25,000 and lists Business at 40,000 against 50,000 for Enterprise. I report the contradiction rather than resolve it. There is no $0 cloud tier, only a 14-day trial.
Where it stops. Self-hosting is licence-free under the Sustainable Use Licence, source-available rather than OSI open source, and not free in any sense that survives a budget line: n8n's own comparison page names servers, monitoring, patches, backups and the person who does all four. At the BLS median of $47.66 an hour for a network administrator, one admin hour a month costs more than Zapier's entry subscription. Our n8n for marketing guide covers the workflow side, and n8n alternatives runs this comparison from n8n's end.
Microsoft Power Automate: $15 a Month for a Ceiling, Not a Pool

Power Automate meters a daily rate, the most important mismatch here. Premium is $15 per user a month by the secondary sources, with 40,000 actions a day per user. Process capacity is $150 a licence a month for 250,000 a day. Five thousand actions a month is about 167 a day, 0.4 percent of one Premium licence.
Where it stops. The ceiling resets, so nothing banks, and a traffic spike meets a hard wall instead of a monthly average. If your flows already live inside Teams, Outlook and SharePoint, the marginal cost of one more is close to zero, an argument no other tool here can make. There is no self-host path.
Activepieces: $20 Flat and the Steepest Per-Credit Drop Here

Activepieces prices flat and scales in steps, the easiest row here to forecast. Plus is $20 a month for 10,000 credits and Team is $200 for 50,000, each tier a multiple of the one below. One credit covers one flow run however many steps it has, against Zapier's one per action. At 5,000 runs a month that is $0.0040 a run, with double your workload inside the allowance.
Where it stops. The one-credit-per-run definition is secondary-sourced; the plan page prints multipliers and never defines the unit. Self-hosting is MIT licensed with no licence fee, and the admin-hour arithmetic above applies unchanged. SSO beyond Team, audit logs, embedding and managed multi-tenant projects sit behind the paid line even when you host it yourself.
Pipedream: $99 for 10,000 Credits You Can Burn Without Writing a Step

Pipedream is the developer row and it prices like one. Startup is $99 a month billed annually, $150 monthly, with 10,000 credits and 100 external users before $2 each. A credit covers an action, a tool call, a Connect proxy request or an emitted trigger event per 30 seconds of compute, and overage is $0.012 a credit after the included 10,000. That marginal rate is the only clean one here. At 5,000 credits the effective rate is $0.0198.
Where it stops. Credits also leave on events you never wrote as a step, so a busy trigger can eat an allowance a step count would not predict. The $0 tier is development only by the vendor's own labelling. Self-hosting does not exist: the docs path returns a 404.
IFTTT: $2.09 for 5,000 Credits That Buy a Different Job

IFTTT is the cheapest number here by a factor of four, and it is not competing for the same job. Starter is $2.09 a month billed annually, $25.08 a year, for 5,000 credits; Pro is $5.49 for 20,000 and Pro+ is $13.99 for 100,000. A credit pays for an Applet run, an MCP call or an AI feature, and every tier carries every feature, so the ladder is credit count, polling speed and support.
Where it stops. An Applet is one trigger and one action, no branching and no code, which fails the moment your workflow needs a second condition. At $0.00042 a credit it is right for consumer-shaped automation and wrong for marketing operations.
The Switching Cost Nobody Prices
Every page answering this question stops at what else exists. None prices what breaks.
The unit remap comes first. A workflow that cost nine units per run is one unit per run on a whole-execution meter, so last quarter's budget describes a different product. Make's credits expire at term end, turning a seasonal peak into a deadline question. Zapier's meter has a ceiling too, and a per-run meter removes it along with the thing that was quietly throttling a runaway loop.
The connector list comes second. Zapier's 9,000 integrations are the product, and every rebuild is a chance to find that your CRM step is first-class on one platform and a generic HTTP request on another. That is where the automation layer touches the rest of your martech stack.
Third, the exemptions. Zapier does not bill Filter, Formatter, Paths or Forms steps; Make does not bill Router or error handlers. Those are worth real money inside a deep workflow and rarely survive a reprice.
What gets rebuilt: the trigger and its authentication, the field mapping on every step, the branching, the error path, the test data that proved it worked. Count units of work, not weeks, because nothing I read measures migration effort and I will not invent a duration. A dozen Zaps is a dozen rebuilds; two hundred client workflows means two systems billing at once for a quarter.
When Staying on Zapier Is Correct
Three situations, and I will defend all of them.
Your workflows are two or three steps. The nine-to-one arithmetic needs depth to bite. A two-step Zap is one task, and one unit on a per-run meter, so the advantage evaporates and you pay a migration to stand still.
The connector is the product. If your flow depends on a niche tool that only Zapier integrates, the cheaper meter is not cheaper, it is unavailable.
Nobody on your team will own a server. Every self-hosted option here converts a subscription into a labour line, and at $47.66 an hour the labour loses to Zapier's cheapest paid tier almost immediately.
Verdict
There is no single winner, and a page that hands you one has skipped the unit question.
Deep workflows at a steady run rate point to n8n, cheapest per run only if somebody already owns the server. Deep workflows that must stay hosted point to Make at $9 for 10,000 credits. A bill you can forecast without watching a meter points to Activepieces at $20 flat for 10,000 credits. Inside Microsoft 365, Power Automate's marginal cost is near zero. If you write code for a living, Pipedream's $0.012-a-credit overage is the most honest marginal price here. If your automation is consumer-shaped, IFTTT at $2.09 for 5,000 credits does that job most cheaply.
If the honest answer to "how many steps is my workflow" is one or two, the correct alternative is the tool you already have.
Frequently Asked Questions
What is the cheapest Zapier alternative?
IFTTT, at $2.09 a month billed annually for 5,000 credits, or $0.00042 a credit. The qualifier is what the credit buys: one Applet run, one trigger and one action, no branching and no code. If your workflow needs a second condition, Make at $9 for 10,000 credits is the real entry point.
Is n8n really free?
Not as a cloud product. There is no $0 cloud tier, only a 14-day trial with 1,000 executions, and Starter begins at 20 EUR a month billed annually for 2,500 executions. The Community Edition is licence-free to self-host, and the costs that come with it are servers, monitoring, patches, backups and the person who does those things. At $47.66 an hour, one admin hour a month costs more than a Zapier Professional subscription.
Can I migrate my Zaps automatically?
I verified no importer for any of the six alternatives here, and I will not claim one exists. What is documented is that these platforms meter different events, so even a clean workflow import would need its unit budget rebuilt. Plan the migration per workflow rather than per tool, and around the connector doing the real work.
Conclusion
The number to carry into the decision is the unit, not the subscription price. Every meter here is defensible. Only one rewards you for building something complicated.
So the honest answer to which Zapier alternative is cheapest is a sentence, not a name: the one whose unit matches the shape of your workflow. Price one run of your busiest workflow on each, multiply by last month's runs, and the shortlist collapses to two tools. Then check who owns a server.
Get Started: Price Your Own Workflow
Write down three numbers before you compare another tool: the steps in your busiest workflow, the runs it does in a month, and whether anyone on your team can be responsible for a server. Then divide the monthly price by the units it actually buys at your real run rate. That number survives a change of vendor.
Sources and dates. All vendor pages read 2026-11-20. Zapier: zapier.com/pricing and zapier.com/blog/what-is-a-task-in-zapier, its 5,000-task price from two secondary pages dated October 2026. Make: its pricing FAQ and make.com/compare/make-vs-n8n, the pricing page having returned HTTP 403. n8n: n8n.io/pricing. Power Automate: learn.microsoft.com, updated 2026-09-22, the $15 and $150 list prices secondary. Activepieces, Pipedream and IFTTT: their own pricing pages. Wage anchor: BLS, May 2025 median for network and computer systems administrators, $47.66 an hour. Every per-unit figure and the nine-to-one arithmetic are my own on those prices.

Price the work, not the plan
See what a month of marketing output costs when the research, the writing and the reporting run in one place, without a second automation subscription underneath. Plans start at EUR 99 a month.


