Zapier: industry default with 7,000+ apps, US cloud, and task-based pricing

What is Zapier?

Zapier is probably the best-known no-code automation platform in the world. Founded in 2011 by Bryan Helmig, Mike Knoop, and Wade Foster in Missouri, USA, the company defined the "SaaS integration without code" category. As of May 2026, more than 7,000 app integrations are available – more than any other platform in the market. That makes Zapier the default for small teams that want to wire two or three apps together quickly.

The core concept is a Zap: a trigger starts the chain, one or several actions process the data, optionally filters, paths, and formatters are added. One action counts as one task. Anyone building a Zap with 5 actions consumes 5 tasks per trigger. The pricing model thus differs fundamentally from Make (operations) and n8n (executions).

The tiers (as of May 2026): Free with 100 tasks/month and 5 Zaps, Starter from USD 19.99/month for 750 tasks and unlimited Zaps, Professional from USD 49/month for 2,000 tasks plus Premium apps, Team from USD 69/month for 50,000 tasks plus multi-user, Enterprise on request for 100,000+ tasks. Zapier bills quarterly or annually; tasks do not roll over to the next month.

Zapier runs exclusively as a US cloud service. An EU region or self-hosting option does not exist – data is processed in US AWS regions. Since 2023 "Zapier AI Actions" with OpenAI and Anthropic modules exist, since 2024 the "Zapier Agents" feature for tool-using agents. AI calls usually cost 2-5 tasks per invocation, depending on model and token volume.

Why it matters

Zapier is the platform with the largest connector catalogue on the market. Anyone using a niche app (Pipedrive variant, industry tool, regional bookkeeping system) most likely finds an official connector in Zapier. The app breadth is the real main reason Zapier still wins in many setups despite high pricing and no EU region.

For a Swiss fiduciary three scenarios make sense. First: quick-win automation. A bookkeeper wants every new invoice from Bexio to land automatically in a Google Drive folder as PDF. Bexio + Google Drive is wired up in Zapier in 5 minutes, costs 1 task per invoice. At 200 invoices/month that is 200 tasks – the Starter tier for USD 19.99 covers it with headroom.

Second: temporary workflows. A marketing stunt, a webinar, a 3-month lead campaign – Zapier is productive in 30 minutes, and after expiry the Zap is simply paused. No server setup, no maintenance, no engineering resource. For one-off or short-term automation, Zapier is often the only sensible choice.

Third: AI agent experiments. With "Zapier Agents", tool-using agents can be built that fire actions across 7,000+ apps. For proof-of-concepts (e.g. "can an agent book my trip on its own?") that is a very low entry barrier. Whether such agents carry production weight is a different question.

The downside is clear: US cloud only, no self-hosting, fast cost escalation on complex workflows. Anyone building an 8-step Zap consumes 8 tasks per trigger; at 1,000 triggers/month that is 8,000 tasks, requiring the Team tier (USD 69). On Make or n8n the same 1,000 workflow executions would be significantly cheaper.

For professional-secrecy data Zapier is not admissible – US CLOUD Act access without EU region is an unfixable weakness. A realistic path is running Zapier for the marketing stack and n8n for client operations in parallel.

How it works

A Zap consists of a trigger and one or more actions. Triggers react to events in connected apps (new mail, new calendar entry, new CRM record, new Stripe charge). Actions execute operations (send mail, create database record, send Slack message, fire an API call). Every step after the trigger is one task – the trigger itself does not count.

Key advanced features: filters (Zap stops if condition not met, no task consumed), paths (multiple parallel routes with different logic, each executed path counts its tasks), formatter (date, text, number transformations, each one task), sub-Zaps (a Zap calls another, counts 2 tasks: one for the call, one for the sub-Zap trigger), webhooks (custom HTTP trigger and action). The Code action allows JavaScript or Python snippets with a 10-second timeout.

A typical AI triage setup: Gmail trigger (new mail) -> Zapier AI Action (classification into 4 topics, costs 2-3 tasks depending on model) -> Paths module (one route per classification) -> one HubSpot action (CRM entry, 1 task) -> Slack action (notification, 1 task). So 5-6 tasks per incoming mail. At 500 mails/month that is 2,500-3,000 tasks, which busts the Professional tier (USD 49 for 2,000 tasks) – the Team tier (USD 69 for 50,000 tasks) is the next step.

Data types are strictly structured. Every trigger and action delivers a defined output schema, available via drag-and-drop in following steps. Built-in formatters cover standard transformations (date, text, numbers, lists). More complex logic requires either a Code action or an external webhook call.

Error handling works via Zap settings: auto-retry on rate-limit errors, auto-replay after service outage, error notifications by mail. True branching on errors (error-trigger workflow like in n8n) does not exist – anyone needing complex error logic builds a second Zap chain with webhook notification.

The multi-user feature (from Team tier) allows shared Zaps, per-user permissions, and audit logs. Connection sharing (org-wide app connections) and SSO are also from the Team tier.

Zapier setup in 5 steps

1. 01Create an account on zapier.com, start with Starter or Professional tier – Free tier only for tests, task limit too tight for production workflows.
2. 02Connect apps: set up OAuth connections for the central apps (Gmail, Slack, HubSpot, Stripe, optionally Bexio).
3. 03Build the first Zap: trigger -> filter -> 2-3 actions. In the editor run a test with real data, check task consumption in the dashboard.
4. 04Activate error notifications: turn on "Notify me when my Zap fails" in Zap settings, plus a Slack/mail alert to operations.
5. 05Set up task monitoring: Zapier dashboard shows tasks per Zap and month – alert at 80% of tier limit, otherwise silent throttling kicks in after overrun.

When to use Zapier

Zapier is the right choice when (a) the apps to connect lie outside the mainstream list (niche tools, industry software), (b) the team should automate without engineering resources, (c) data protection with the US cloud is acceptable (non-sensitive data), and (d) task volume stays below 2,000-5,000/month.

Concrete cases: simple 1-to-1 app links (Stripe charge -> bookkeeping, Eventbrite signup -> CRM, Google Form -> Slack notification), niche-app wirings (regional bookkeeping -> cloud storage, industry CRM -> mail automation), marketing stunts and temporary campaigns, quick-win automation for individual staff without IT support.

For teams without DevOps capacity, Zapier is often the pragmatic entry. CHF 20-50/month for 1-3 production Zaps, no server, no risk. When the setup grows (10+ Zaps, 5,000+ tasks/month, more sensitive data), migration to self-hosted n8n is the next step – by then the team has learned through Zapier which workflows actually carry value.

For Zapier AI Action experiments: a fast lane to LLM-driven workflows without API-key management. The AI Action calls OpenAI or Anthropic in the background, token cost is included in the task price – that simplifies the calculation but is markedly more expensive per token than direct API access.

When not to use

Zapier is wrong for professional-secrecy data. US cloud without EU region and without self-hosting is a clear no for fiduciary and legal data – even with Standard Contractual Clauses, CLOUD Act access remains. Client mail, tax data, payroll data do not belong on Zapier.

Zapier is unsuited at high task volume with complex logic. Anyone needing 50,000+ tasks/month pays Team or Enterprise tiers (USD 69-500+/month) – for that money, self-hosted n8n plus engineering effort is fundable. Eight-step Zaps multiply the problem: 1,000 triggers * 8 tasks = 8,000 tasks/month, the cost curve explodes.

Zapier does not fit complex workflow logic. Nested branching, iteration over large lists, reusable sub-workflows – the models in Zapier (Paths, Sub-Zaps) are less powerful than router/iterator/aggregator in Make or the sub-workflow architecture in n8n. Complex setups grow unwieldy quickly.

Unsuited for data pipelines (ETL, data warehouse, cleansing): per-task pricing makes processing large datasets unaffordable. Airflow, dbt, Airbyte are the right tools there.

Not suited for AI agent logic with many tool calls and reflection: Zapier Agents are a good proof-of-concept mechanism, but for production agent systems with long state and many iterations LangGraph, AutoGen, or CrewAI are more fitting.

Trade-offs

FAQ

Related topics

Sources

1. Zapier documentation – Zaps, tasks, AI Actions · 2026-05
2. Zapier pricing page · 2026-05
3. Zapier Agents and AI features overview · 2026-04
4. Zapier security and data processing overview · 2026-03