Zapier vs Make 2026: They Don't Bill for the Same Thing

Every Zapier vs Make comparison, including the first version of this one, puts $19.99 for 750 tasks next to $9 for 10,000 credits — both prices billed annually — and lets you draw the obvious conclusion: thirteen times more, for half the price.

We went and read both companies' billing documentation. Those are not the same unit, they are not close to the same unit, and the difference runs in a direction the headline comparison completely hides.

Our 60-day test still stands and nothing measured in it has changed. What follows is that test, plus the thing it was missing.

How We Tested

Plans: Zapier Professional and Make Core, both paid, both running simultaneously. Duration: 60 days. Workload: the same 10 automations on each platform, at the same volume — roughly 500 form submissions, 200 client follow-up sequences, 150 calendar events synced and 100 invoice notifications a month, plus smaller flows. Measured: what each platform cost at that volume, whether every automation could be built on each, how long the first automation took to build, how long until we felt competent, and how many runs failed.

Two results worth stating up front because neither flatters our conclusion:

• Zapier had zero failed runs in 60 days. Not one. If reliability is what you're buying, we didn't catch Zapier failing
• Make took about a week before we were comfortable, against five minutes for a first Zapier automation. That's a real cost, paid in your time, and it doesn't appear on either invoice

What this test does not cover:

• Which triggers were polling and which were webhooks. We didn't record it, and the research below shows that single variable can move Make's number by an order of magnitude. This is a real gap and it's new — we only know it matters because of the billing work in this update
• Long-run reliability. Sixty days with no Zapier failures is not the same as proving Make fails more; we didn't record enough failures on either to compare
• Support quality. We didn't open tickets with either vendor
• The full app catalogues. We found two tools in our own stack that Make didn't support directly. Which two matters entirely to us and not at all to you — check your own
• Volume tiers above ours. Both platforms price in bands; our figures are the bands our workload landed in

Zapier bills for work completed

From Zapier's own help centre: "A task is any successful action that runs in Zapier. Only successful actions count toward your task usage."

Here is the list of things that consume nothing:

• Trigger steps
• Filter steps
• Paths steps
• Failed or halted actions
• Steps skipped because of a filter or an error
• Zapier Tables and Forms steps
• Formatter, Delay, Looping, Digest, Zapier Manager and Storage

And the line that matters most, quoted exactly:

"Polling triggers will poll an app or check for new trigger events at a regular interval, but these polls never use tasks."

So a five-step Zap — trigger, filter, three actions — costs three tasks per run. The trigger is free, the filter is free, and if the filter stops the run, the actions you didn't reach are free too.

A few 2026 specifics from the same document: MCP tool calls cost 2 tasks each, Lead Router costs 5 tasks per routed lead, and Zapier Agents don't consume tasks at all — they run on a separate activity quota.

On overage: with pay-per-task on, extra tasks bill at 2.5× the base rate monthly or 1.25× annually. With it off, everything stops until the next cycle. Either way there's a hard ceiling at 3× your subscription.

Make bills for the machine running

Make's model is the mirror image. Every module call consumes a credit — including the trigger, and including each filter evaluation. Router branches execute their modules. Make's Code modules bill at 2 credits per second of execution.

The same five-step workflow that costs Zapier three tasks costs Make five credits.

That alone would be manageable. This is the part that isn't:

A polling trigger spends a credit every time it checks, whether or not anything has happened.

Work it through. Make Core allows a one-minute minimum interval. A single scenario polling every minute checks 43,200 times a month — over four times Core's entire 10,000-credit allowance, consumed before a single piece of real work happens. At five-minute polling it's 8,640 checks a month, roughly 86% of Core, to watch a folder that might receive a file once a week.

On the free plan it's faster: at the 15-minute minimum, a four-module scenario burns through the 1,000-credit allowance in under three days.

Two more things multiply it. Iteration compounds: one iterator processing 20 items through 5 modules is 100 credits from a single trigger event — a tenth of the free plan, in one run. And dormant scenarios still cost. In one migration case we read about, an agency moving 127 Zaps found 34 of them dormant — switched on, doing nothing. On Zapier a dormant Zap with a polling trigger costs exactly zero. On Make it had been spending credits the entire time.

A webhook trigger consumes nothing while it waits, and one credit when a real event arrives.

That is the single most valuable sentence in this article. Converting one polling trigger to a webhook can save more than the entire price difference between the two platforms — and it appears in no Zapier vs Make comparison we read.

A note on sourcing. Zapier's polling line is quoted directly from its help centre. Make's credits documentation describes the unit without addressing polling specifically; the behaviour is documented in Make's Academy material and corroborated by multiple independent practitioners with consistent arithmetic, and contradicted by nobody. We're telling you where each fact comes from because the difference matters.

So what does the price comparison actually mean?

Not what everyone says it means.

Zapier charges you for work completed. Make charges you for work attempted. Which is cheaper depends entirely on how much of your automation's activity is checking versus doing — and that is a property of how you built it, not of the price list.

If your scenarios are webhook-driven and your workflows are long, Make is dramatically cheaper and the 13x-for-half-the-price framing roughly holds. If you have several polling triggers watching things that rarely change, Make can cost more than Zapier while appearing to cost a fifth as much.

So the useful habit is not choosing a platform. It is auditing what you have switched on — because on Make, an automation you forgot about is a line item.

There aren't two billing units. There are five.

Once you look at the unit rather than the price, the whole category reorganises:

n8n's own pricing page: "An execution is a single run of your entire workflow. It doesn't matter how many steps are in the workflow or how much data it processes." Pipedream's docs: "one credit per 30 seconds of compute time at 256MB" and "Pipedream does not charge for usage based on the number of steps" — but doubling memory doubles the cost. Activepieces moved off task-based pricing entirely: free to 10 active flows, then $5 per flow per month with unlimited runs.

The same workflow is cheapest on a different platform depending only on its shape. Many steps and few runs favours n8n or Pipedream. Few steps and many runs makes Zapier reasonable. Constant polling punishes Make and costs Zapier nothing. Enormous run volume across a handful of flows makes Activepieces nearly free.

This is the same pattern we found looking at Make alternatives: the competitors changed the billing unit, not just the number.

Our test, re-read

At our volume the measurement was ~2,500 tasks on Zapier (the $49 Professional band) and ~3,000 credits on Make ($9 Core, of 10,000 included).

Corrected 10 September 2026 — those two figures don't fit together. On Zapier's pricing page, read with its task selector and both billing options, $49 billed annually buys 2,000 tasks. About 2,500 tasks needs the 5,000-task band: $89 billed annually, or $133.50 monthly. Our records can't tell us which was wrong — the task count or the band — so here are both readings. If the band was right, our volume was nearer 2,000 tasks. If the count was right, the gap at our volume was about $80 a month on annual billing, not $40. And Make's $9 is Core billed annually; month to month it's $10.59.

The unit analysis explains something we didn't notice at the time. If Make charges for triggers and filters and Zapier doesn't, then consumption that close is only consistent with our scenarios being mostly webhook-driven or infrequently polled. Had we polled aggressively, the Make figure would have been vastly higher and the conclusion could have inverted.

We didn't record which triggers were which. That's now in the limits list above. It doesn't undermine what we measured — it bounds what the measurement can be used for, and we'd rather say so.

On the $480/year figure: at our volume we recorded $49 against $9 — a $40/month gap, or about $80 if our task count was right and the band wasn't (see above). Either way it's a band our workload landed in, not Zapier's entry price. Zapier Professional starts at $19.99/month billed annually ($29.99 monthly) for 750 tasks. If your volume fits inside 750 tasks, the gap is about $11/month billed annually, or $19.40 monthly, not $40.

The bit that's new in 2026: AI assistants are now the client

If you connect Claude or ChatGPT to your apps through Zapier MCP, you're spending from the same meter — at double rate.

Quoted from Zapier's documentation: "Each successful tool call through your MCP server consumes two tasks," and "Tasks used through MCP count toward your plan's total task allowance."

So the Free plan's 100 tasks is roughly 50 AI tool calls a month, shared with your Zaps. Professional's 750 is about 375. Asking your assistant to find a contact and email them is two tool calls — four tasks. Adding five spreadsheet rows is five calls, ten tasks.

Failed calls don't count, and neither does authentication or setup. And Zapier Agents don't touch tasks at all — they draw on a separate activity quota.

We couldn't pin down an equivalent cost model in Make's own documentation, so we're not going to guess at one.

Most comparisons get the unit wrong — including the ones an AI will read

This is worth knowing because it's why the received wisdom is what it is.

Of the comparison articles we opened, three state that Zapier bills for filters and formatters — which Zapier's own help centre explicitly denies. One says a five-step workflow costs five tasks, in the same passage where it correctly says triggers are free. Another puts Zapier at "over 6,000 apps" when Zapier's own catalogue page says 10,057+.

The two that got it right both cited Zapier's official task-counting article. The accurate ones went to the primary source. That is the entire difference, and you can verify it yourself in about a minute.

The vendors are no better on each other. Zapier's blog about Make correctly notes Make charges for triggers, filters and routers — and frames its own exclusions as a virtue. Make's blog about Zapier correctly notes the $9-vs-$19.99 gap — and omits that credits and tasks measure different things, which is the fact that makes the comparison meaningless as stated. Each is accurate about its own billing and misleading about the comparison.

App counts are not a usable comparison

Our previous version said Zapier connects "over 9,000 apps." Checking on 4 September 2026:

• Zapier's own apps page: "10057+ apps"
• Zapier's Copilot and MCP pages: "9,000+"
• Zapier's enterprise material: "8,000+"
• Make's own integrations page: "3,000+"
• Third parties: Zapier anywhere from 6,000 to 10,000; Make at 2,000 or 3,000

Zapier's own pages disagree with each other by two thousand apps. Treat every headline integration count as marketing. The only number that matters is whether your apps are supported, which takes two minutes to check on either site.

Head-to-Head

Which one should you choose?

Choose Zapier if you've never automated anything, your workflows are short, you need an app Make doesn't support, or — the new reason — several of your automations poll things that rarely change. Zapier doesn't charge for waiting.

Choose Make if you're comfortable with a learning curve, your logic branches, your workflows are long, and your triggers are webhooks rather than polls. That combination is where Make's price advantage is real rather than theoretical.

And the free-plan advice inverts. If you're paying from day one, start with Zapier — the five-minute first automation is real, and so is Make's week-long learning curve. If you're starting free, start with Make: Zapier Free allows one action per Zap, and in our separate 30-day test of both free tiers (Zapier Free vs Make Free) Zapier covered 2 of 10 workflows against Make's 8.

On switching later: we haven't migrated between them, so we won't tell you how long it takes. Zaps can't be imported into Make — scenarios have to be rebuilt. Reported experience ranges from a few hours for a handful of automations to several weeks for a hundred-plus. For a solopreneur running five to ten, expect it to be an afternoon or two of real work rather than a click.

For setup, our beginner Make tutorial builds five automations and explains which two to leave running on the free plan. For the wider stack, see the 2026 solopreneur AI stack.

---

Both tools were tested for 60 days running the same 10 automations at the same volume, on Zapier Professional and Make Core. Method and limits are set out under "How We Tested" above, including one limit added on 4 September 2026 that we did not know to record at the time. Billing definitions quoted from Zapier's and Make's own documentation, and prices re-verified against both vendors' pricing pages, on 4 September 2026. Prices were read again on 10 September 2026 with both billing options and Zapier's task selector, which is how we found the $49 band inconsistent with our own recorded task count. Links marked "Paid link" earn us a commission if you sign up through them, at no extra cost to you; no other link here pays us — how we make money.