Dimitri Cabete Jorge, Co-founder & CTO · Last updated: August 2026 · Facts verified: August 2026
TL;DR verdict: Zapier is the easiest place to start with no-code automation, and it connects to more apps than anything else in the category. The catch is how it charges. You pay for every action step every time a workflow runs, so the bill grows with the number of steps and how often they fire, and the free and entry plans run out sooner than most buyers expect. It suits non-technical teams on common SaaS tools who need wide app coverage more than complex logic. If your workflows are long, high-volume, or full of branches, look at Make before you commit.
Ratings: G2: 4.5/5 (1,818 reviews) · Capterra: 4.7/5 (3,048 reviews), verified August 2026
What Is Zapier?
Zapier is a no-code automation platform that moves data between web apps and runs multi-step workflows for you. Its workflows are called Zaps. Each Zap has one trigger, the event that starts it, followed by one or more actions, the things Zapier does once it fires.
Zapier now markets itself as an AI orchestration platform, and the 2026 product line matches that pitch. It ships AI agents that work out their own steps, chatbots, and a server that lets outside AI tools such as Claude and ChatGPT act inside your apps. The underlying product has not changed category. It is still integration software, the kind you configure in a visual builder instead of writing code. By its own count, 2 million businesses run 81 billion tasks across 9,000+ apps, and 69% of the Fortune 1000 are customers.
What Is Zapier Used For?
Zapier is used to hand repetitive work between apps so nobody has to retype it. That covers copying a record from one tool into another, alerting the right person, and now letting an AI agent run several of those steps on its own. The examples below are the patterns teams build most often.
- Lead capture and CRM routing: A form submission creates a CRM record, posts a Slack alert, and adds the contact to a newsletter list without anyone retyping the data.
- AI lead qualification: Call recordings get transcribed, an AI step scores lead quality and extracts the prospect's name and timeline, and the result lands in the CRM.
- Support ticket pre-population: ClickUp's support team cut per-ticket research from 15 minutes to 4 across 5,000+ monthly tickets, or roughly 917 hours monthly.
- Support routing: Hudl saves $12,000 to $15,000 a year and cut average handle time by 21.5% after automating how requests get assigned.
- Invoice confirmation: A payment event in an accounting tool fires the confirmation email, so nobody chases it per transaction.
- Internal IT and HR requests: Teams wire Zapier into access requests, onboarding checklists, and offboarding steps, the same recurring HR tasks and IT handoffs that cross departments and generate the longest approval chains.
- AI agents and MCP: Agents run multi-step work across connected apps on their own, and Zapier's MCP server lets tools like Claude and ChatGPT fire 30,000+ actions without a pre-built Zap.
Every one of those patterns moves a record from A to B, and the bill tracks how many times it does it.
Key Features of Zapier
Zapier's feature set now comes in two halves: the original Zap builder, and an AI layer added through 2025 and 2026. These are the parts that should change a buying decision.
- Zaps (core workflow engine): A Zap is a trigger plus one or more actions. Filters restrict runs to conditions you set, and Paths add if/then branching on Professional and above. Multi-step Zaps are paid-only.
- 9,000+ app integrations: The broadest connector library on the market, and the main reason teams start here.
- Zapier Agents: With a Zap, you build the steps and Zapier runs them. With an agent, you describe the outcome and the agent works out the steps itself, using your instructions and live data from the same connected apps. Nothing has to be built in advance. In February 2026 Zapier added AI Guardrails to screen AI output for safety and compliance.
- Zapier MCP (Model Context Protocol): MCP is a standard that lets an AI assistant do things in your apps rather than only talk about them. Zapier's MCP server gives an AI tool 30,000+ ready-made actions and handles the logins and rate limits for you. More than 195,000 of these servers have been created. Each action an AI tool takes this way costs 2 tasks, drawn from the same allowance as everything else.
- Zapier Copilot: An AI assistant available on login and inside every Zapier product, covering Zap workflows, Forms, Tables, Agents, and Chatbots. It can produce a working Zap, a table, or a form in one chat session, and diagram a process in Canvas before you build it. Copilot is unlimited on paid plans and capped daily on Free.
- AI by Zapier: Built-in access to large language models with a choice of provider and model, plus knowledge sources pulled from Google Drive, Notion, and Confluence. Steps have been priced by model tier since June 15, 2026. Available on Professional, Team, and Enterprise.
- Code by Zapier: Custom Python or JavaScript inside a Zap, with npm and PyPI package support and included runtime capped at 1 second on Free, 30 seconds on Professional and Team, and 2 minutes on Enterprise. Zapier Functions is being retired in September 2026, on the 1st, and new functions have not been available since June 1, 2026, so anything running there needs moving to Code by Zapier before the cutoff.
- Tables, Interfaces, and Canvas: A spreadsheet-style store for automation data, brandable forms that trigger workflows from responses, and an AI diagramming surface for mapping a process before you build it. Tables and Forms steps consume no tasks on any plan.
- Zapier Chatbots: AI bots for internal or external questions, embeddable on a site or reachable from email, able to answer from company knowledge and fire a Zap.
- Developer SDK: A native way for code-environment builders to add Zapier's connector catalog to an agent project. SDK usage draws on the same shared task pool as Zaps do.
One pattern runs through that list. Zapier has spent two years adding new ways to spend tasks without changing how tasks are counted.
Zapier Pros & Cons
Zapier is easy to start with and expensive to scale.
What Users Say
Users rate Zapier highly on the software review sites and badly on Trustpilot. That gap is the most useful thing in the numbers, because the two audiences are not complaining about the same product.
What users consistently praise:
- Breadth of pre-built connectors is the clearest competitive advantage, and it shows up on every platform.
- Non-technical users describe the trigger-action builder as fast to learn, and a simple Zap really is a few minutes of work.
- Simple, linear automations against common SaaS tools run dependably without ongoing maintenance.
- Clients and agencies often ask for Zapier by name, which ends the argument about which tool to use before it starts.
- Copilot earns credit for cutting setup time by generating a starting Zap from a plain-language description.
What users consistently complain about:
- Cost is the single most-cited complaint theme, and a second cluster ties the surge specifically to advanced, multi-step workflows.
- The free and entry tiers run out faster than people plan for, particularly once a workflow has more than one action.
- Debugging is the next consistent theme, and the specific pain is diagnosing a multi-step failure from logs written for someone who already understands the platform.
- A recurring set of reviews describes Zaps that stopped working with no notification. That is the version of the reliability complaint that loses data rather than just wasting time.
- Support is described as email-only on most tiers, slow, and prone to closing tickets before the problem is resolved.
- The free plan's feature limits draw their own complaints, separate from the task cap.
Where reviewers diverge:
- Buyer platforms and Trustpilot capture different populations. Buyer platforms skew toward people evaluating automation capability; Trustpilot skews toward people who have just been surprised by a charge.
- The Trustpilot complaint pattern is narrow and specific: unexpected charges, subscription changes without prior notification, and difficulty cancelling. That is a billing verdict, not a product verdict.
- Ratings follow how heavily someone uses the product, not how good they think it is. The same person who praises the builder at three Zaps complains about the bill at thirty.
None of the complaint themes are about whether Zapier works. They are about what it costs once it does.
Zapier Pricing
Zapier bills by task, and a task is one successful action step. Triggers are free. So are filters, paths, formatters, delays, and the looping step. The number you pay for is therefore smaller than the number of steps on your screen.
That definition does most of the work in a budget. A Zap with four action steps firing 200 times a month consumes 800 tasks. Professional's base allowance is 750, so that one workflow uses up the whole plan. Every other Zap on the account then costs extra. Run the same Zap 50 times a month and it costs 200 tasks, comfortably inside the allowance.
The variable that decides your bill is not how many Zaps you have. It is action steps multiplied by runs. That number is also the only honest basis for measuring automation payback later.
Rates come from zapier.com/pricing. All pricing information verified August 2026.
Gotchas:
- Monthly billing costs 50% more: The advertised $19.99 and $69 are annual-billing rates. Pay monthly and Professional is $29.99, Team is $103.50. Zapier presents this as a 33% saving for paying yearly, which is the same fact counted from the cheaper end: $29.99 less a third is $19.99, and $19.99 plus half again is $29.99.
- Buying more tasks barely earns a discount at the low end: Above the 750-task base, Professional costs $39 a month billed annually for 1,500 tasks, $49 for 2,000, and $89 for 5,000. Zapier says the price per task falls as the tier rises, and it does, but mostly higher up the ladder. Between 750 and 2,000 tasks it is close to flat, so doubling your allowance roughly doubles your bill.
- Live chat is gated by volume, not by plan: Email support starts on Professional, but live chat only begins at Professional's 2,000-task tier, which is $49 a month billed annually.
- Free plan restrictions are real: Two-step Zaps only, 15-minute polling, no premium apps, no webhooks, no autoreplay. Tables cap at 2,500 records per account, Copilot has a daily message limit, and Code by Zapier gets 1 second of included runtime.
- Multi-step Zaps are paid-only: Two steps is a hard wall for anything past a single trigger and a single action.
- Polling intervals by plan: 15 minutes on Free, 2 minutes on Professional, 1 minute on Team and Enterprise. Time-sensitive automations need Professional at minimum.
- One shared task pool: Zap workflows, AI steps, Code, MCP, and SDK usage all draw on the same account-wide allowance. Agents are the single exception and meter separately.
- AI tool calls cost double: When an AI tool acts through MCP, each successful action bills 2 tasks instead of the 1 a normal action step costs. Failed calls cost nothing.
- Lead Router costs 5 tasks per lead: One routed lead consumes five times what a standard action step does.
- AI by Zapier is priced by model tier: Standard costs 1 task per run, Advanced 3, and Premium 5, with tool calls billed at the same multiple. New steps default to Advanced, so 3x is the rate you get unless you change it. A single step that runs up 75 tasks pauses the Zap and asks for approval before continuing.
- Overage is on by default: A paid plan that reaches its allowance does not stop. It carries on, and the tasks beyond the allowance bill separately at a worse unit rate. Switching that behaviour off is a setting, not a default, and it changes the failure mode rather than removing it.
- Zapier contradicts itself on Team seats: The Team plan card and feature table both say 25 users, while the FAQ further down the same pricing page says Team and Enterprise allow unlimited users. Confirm the seat count in writing before signing.
- Free trial: New accounts get 14 days of Professional with no credit card required.
- Zapier Functions ends September 1, 2026: New functions have not been available since June 1, 2026. Existing code is preserved, but it needs moving to Code by Zapier before the cutoff date.
- Agents and Chatbots are tiered above a free allowance: Agents Free covers 400 activities a month, Agents Pro is $400 billed annually ($33.33 a month) for 1,500, and Agents Enterprise is quote-only. Chatbots Free builds two bots, Chatbots Pro is $160 billed annually ($13.33 a month) for five, and Chatbots Advanced is $800 billed annually ($66.67 a month) for 20 with 30 days of conversation history.
Is Zapier Worth It?
Yes, if you are a non-technical team automating straightforward work across common SaaS tools. No competitor connects to as many apps, Copilot cuts the setup time, and the one thing reviewers agree on across every platform is that simple Zaps run reliably. The 2026 AI features also make Zapier a sensible first AI automation tool for a team with no engineering time to spare.
Look elsewhere if your workflows are high-volume, loop-heavy, or multi-branch, because that is where per-task billing stops being a rounding error. Make is built for that shape of work. Look elsewhere too if you have data residency requirements or cannot run automation in someone else's cloud, which is n8n's whole argument.
Zapier vs Make
The two tools count your usage differently, so their headline prices are not comparable. Zapier charges for successful action steps only. Triggers, filters, paths, formatters, and loops are free. Make charges one credit for every step in a scenario, the trigger and the filters included.
The way to compare them is to price the same workflow on both. Take a workflow with one trigger, one filter, and three actions.
On Zapier that costs 3 tasks per run, because the trigger and the filter are free. Professional's 750 tasks buys about 250 runs a month. On Make the same workflow costs 5 credits per run, because everything counts, but its entry paid tier includes 10,000 credits. That buys about 2,000 runs. Make counts more of the workflow against you and still ends up with roughly eight times as many runs.
That gap narrows as workflows get shorter, because more of a short Zapier workflow is made of free steps. It widens as workflows get longer, because Zapier's action count is what grows.
Depth runs the other way. Make's steps expose more of each app's features and give finer control over how data moves between them. Zapier's catalog is wider but shallower per app.
Where Zapier wins:
- App coverage: 9,000+ apps against Make's 3,000+.
- Learning curve: The linear builder is faster for a non-technical person to pick up than a visual canvas of branching modules.
- Free steps: Triggers, filters, paths, and formatters consume no tasks, while Make counts every module call including the trigger.
- Agents: Zapier's agents can act across your connected apps without a workflow being built first.
Where Make wins:
- Allowance size: 10,000 credits on the entry paid tier against Zapier Professional's 750 tasks.
- Branching and loops: The visual canvas is built for the non-linear logic that Zapier's linear structure makes awkward.
- Scheduling precision: Make's Core tier schedules down to a one-minute interval, ahead of Zapier Professional's 2-minute polling.
See the head-to-head breakdown.
Zapier Alternatives
Most teams who leave Zapier go to one of two places, and which one depends on why they left. Cost at volume sends them to Make. A need to control their own infrastructure sends them to n8n.
Make: Pick this if your workflows involve branching paths, loops, or task volumes where per-task billing has become a line item you have to defend. Switching cost is low, because Make is still a no-code visual builder and most Zaps map onto scenarios without a rebuild. Price your actual workflows before you move. Make bills for steps Zapier gives away free, so the saving is smaller on short workflows than the allowance sizes suggest. Make leads the broader replacement field for that reason, ahead of the enterprise-governance and self-hosting options.
n8n: Pick this if you need self-hosted automation with full data control and no dependence on someone else's cloud. The self-hosted Community Edition is free with unlimited workflows, users, and steps, and the hosted Starter plan runs €20 a month billed annually for 2,500 workflow executions, priced per execution regardless of how many steps a workflow contains. The trade is engineering time, because n8n expects someone comfortable with infrastructure, which is why it lands with DevOps teams rather than marketing ones.
How Zapier Works With Siit
Siit does not integrate directly with Zapier.
Teams point Zapier at internal IT and HR work because those requests cross departments, and cross-department requests need approvals. Approvals are what make a workflow long, and length is what per-task billing charges for. An access request that touches a manager, an IT owner, and an identity provider is already a five- or six-action Zap. Compliant approval workflows only get longer as headcount grows.
Siit handles that whole request instead of automating the steps inside it. Its catalog runs to 500+ connectable apps, and requests come in through Slack or Microsoft Teams rather than a portal nobody logs into. Employee data syncs in from HR systems like BambooHR and Workday. That sync is read-only, so the HR system stays the owner of the record.
Identity work runs through the identity providers Siit documents, and each one supports a different set of actions. Okta is the deepest. From a request side panel, a workflow, or an IT Agent playbook, an agent can:
- Reset a password or reset MFA
- Suspend, activate, or deprovision a user
- Add a user to a group or remove them from one
- Assign an application to a user
- Revoke a user's active sessions
Any of those can be gated behind an approval. JumpCloud covers the same ground apart from the application actions. Google Workspace is deliberately narrower: it syncs the directory and adds or removes users from Google Groups, and has no user lifecycle actions of its own.
Device actions are narrower, and worth stating plainly. An agent can lock a device, wipe it, or open its record in the management console, through Jamf, Microsoft Intune, or Iru (formerly Kandji). Those three commands plus device inventory sync are the whole set. There is no restart or shutdown, and disk encryption keys stay in the device management tool.
The difference shows up at the point the request arrives. Where Zapier sees a cross-department request as a sequence of billable steps, Siit's employee-facing intake treats it as a single request that carries its own approvals.
Siit's AI agents handle triage, routing, and answering from your knowledge base. They do not send device commands. If most of your Zapier bill comes from internal approval chains, autonomous request resolution is worth pricing against it. The coordination happens inside a single request rather than as a metered sequence of steps.
