n8n vs Zapier vs Make: Cost, Privacy & GDPR (2026)

You need workflow automation. You have shortlisted n8n, Zapier, and Make. Now you need hard numbers on cost and clear answers on where your data goes.

This comparison gives you real pricing math with specific workflow scenarios. Not pricing tier screenshots. Not a sales pitch for any of the three platforms.

We cover what matters for the decision: how each platform charges you, what it actually costs at different scales, where your data lives, and what GDPR compliance looks like in practice. At the end, a decision matrix to match your situation.

How does n8n pricing compare to Zapier and Make in 2026?

Each platform uses a different billing unit. This single difference drives most of the cost gap at scale. n8n charges per workflow execution. Zapier charges per task (each action step). Make charges per credit. Until August 2025, Make called these "operations." The balance conversion was 1:1, but credits now also cover AI modules and advanced features that can consume more than one credit per action. Check Make's credits documentation for the current breakdown.

What does a workflow execution actually cost on each platform?

A workflow execution on n8n counts as one execution regardless of how many nodes it contains. A 3-node workflow and a 50-node workflow both cost 1 execution. On Zapier, every action step in a Zap counts as one task. A 5-step Zap running once burns 5 tasks. On Make, each module that processes data consumes at least 1 credit. A scenario with 8 standard modules running once uses 8 credits.

This means a 5-step workflow running 100 times per day costs:

n8n: 100 executions/day = 3,000/month
Zapier: 500 tasks/day = 15,000/month
Make: 500+ credits/day = 15,000+/month

The multiplier effect is why n8n gets cheaper as workflows grow in complexity.

How the billing models compare

	n8n Self-Hosted	n8n Cloud	Zapier	Make
Billing unit	Execution	Execution	Task	Credit
What counts	1 workflow run = 1 unit	1 workflow run = 1 unit	Each action step = 1 unit	Each module run = 1+ units
Free tier	Unlimited (Community Edition)	Yes (limited)	Yes (limited)	Yes (limited)
Pricing model	Infrastructure cost only	Tiered monthly plans	Tiered monthly plans	Tiered monthly plans
Paid plans scale by	Server resources	Execution volume	Task volume	Credit volume

For current plan prices: n8n pricing, Zapier pricing, Make pricing.

Is self-hosted n8n really free?

The n8n software itself costs nothing to run on your own server under the Sustainable Use License. This is a fair-code license, not open source. You can use it for internal business purposes and personal use. You cannot resell it or distribute it commercially. The Community Edition gives you unlimited executions, all 400+ integrations, and no per-workflow limits.

But "free software" does not mean "free to operate." Self-hosting costs real money:

True self-hosting cost breakdown

Component	Monthly cost	Notes
VPS (2 GB RAM, 1 vCPU)	12 EUR	Enough for light workloads. Virtua.Cloud vCS-2 with SSD and DDoS protection
VPS (4 GB RAM, 2 vCPU)	24 EUR	Recommended for 50+ active workflows
PostgreSQL	0 EUR	Runs on same VPS via Docker. Included in server cost
Backups	0-5 EUR	Automated VPS snapshots or `pg_dump` to object storage
Domain + SSL	0 EUR	Let's Encrypt for TLS. Domain you likely already own
Your time (maintenance)	Variable	Updates, monitoring, troubleshooting. Estimate 1-2 hours/month
Total (light)	12-17 EUR	Solo indie hacker, <20 workflows
Total (moderate)	24-29 EUR	Small team, 50-100 workflows

The tradeoff is your time vs paying for a managed cloud plan. At any significant volume, the per-execution billing model of self-hosted n8n (where executions are effectively free) creates a widening cost gap compared to per-task or per-credit billing.

Why self-hosting gets cheaper at scale

The billing model difference compounds as you scale. Consider three scenarios:

Scenario 1: Solo indie hacker, 10 workflows, ~3,000 executions/month

Each workflow averages 4 steps. That means 3,000 executions on n8n, but 12,000 tasks on Zapier and 12,000 credits on Make. Self-hosted n8n on a 12 EUR/mo VPS handles this with room to spare. Check each platform's pricing page to see which SaaS tier covers 12,000 tasks or credits.

Scenario 2: Small team, 40 workflows, ~15,000 executions/month

Average 5 steps per workflow = 75,000 tasks on Zapier, 75,000 credits on Make. Self-hosted n8n on a 24 EUR/mo VPS (4 GB) handles this. The SaaS platforms need higher-tier plans to cover 75,000 units. The cost gap grows to roughly 5-10x depending on the plan.

Scenario 3: High-volume automation, 100+ workflows, ~50,000 executions/month

Average 6 steps = 300,000 tasks on Zapier, 300,000 credits on Make. Self-hosted n8n on a 48 EUR/mo VPS (8 GB) handles this. At this volume, you are looking at enterprise-tier SaaS pricing. Users on X regularly report switching from high Zapier bills to self-hosted n8n at a fraction of the cost.

Hidden costs people forget

The scenarios above cover the obvious costs. Here are the ones that catch people off guard:

Zapier task overages. Exceed your plan's task limit and Zapier charges for extra tasks at a higher per-task rate. A busy month can double your bill.
Make credit burn on errors. Failed module executions still consume credits. A misconfigured scenario retrying in a loop can drain your monthly allowance in hours.
n8n disk space. Execution logs with large payloads (file attachments, API responses) fill up your VPS disk. Monitor disk usage and configure EXECUTIONS_DATA_MAX_AGE early.
n8n update labor. Major version upgrades (like n8n 1.x to 2.x) sometimes require migration steps. Budget 2-4 hours per major upgrade.
Webhook costs on all platforms. External services calling your webhooks at high frequency count against your execution/task/credit limits. Rate-limit your webhook endpoints.

Self-hosted n8n keeps all workflow data, credentials, and execution logs on your server. Nothing leaves your infrastructure. For businesses processing EU personal data, this is the simplest path to GDPR data residency compliance. But "self-host = GDPR compliant" is an oversimplification. You still have responsibilities.

Where does each platform store your workflow data?

	n8n Self-Hosted	n8n Cloud	Zapier	Make
Data location	Your server (you choose)	EU (Azure Frankfurt, encrypted at rest)	US (AWS) by default	EU or US (AWS), selectable at signup
EU data residency	Yes (host on EU VPS)	Yes (EU by default)	Optional (enterprise feature)	Yes (eu1.make.com)
Self-host option	Yes	No	No	No
DPA available	N/A (you are the controller)	Yes (includes SCCs)	Yes (includes SCCs, EU-US DPF)	Yes (includes SCCs, EU-US DPF)
Encryption at rest	You configure it	AES-256 (Azure SSE)	Yes (AWS)	Yes (AWS)
Credentials storage	Your PostgreSQL DB	Encrypted in n8n Cloud	Zapier vault	Make vault

Zapier stores data in the US by default. For EU data residency, you need their enterprise plan. Make lets you choose EU or US hosting at signup (eu1.make.com for EU). No enterprise plan required for EU data residency on Make. n8n Cloud runs on Azure in Frankfurt, Germany.

Choosing self-hosted n8n gives you full control. It also gives you full responsibility.

As data controller (self-hosted n8n), you must:

Prune execution logs. n8n stores execution data including input/output payloads. If workflows process personal data, configure log retention in n8n settings or set up a cron job to clean old executions from PostgreSQL. Set EXECUTIONS_DATA_MAX_AGE in your environment.
Manage encryption keys. The database stores workflow credentials. Encrypt the PostgreSQL volume. Use a strong N8N_ENCRYPTION_KEY environment variable (generate with openssl rand -hex 32).
Handle data subject requests. If someone requests data deletion (GDPR Article 17), you need to find and remove their data from execution logs manually.
Maintain access controls. Restrict who can access the n8n instance. Use reverse proxy authentication, strong passwords, and IP allowlisting.

With Zapier or Make (SaaS), they handle:

Infrastructure encryption and patching
DPA compliance (Data Processing Agreement)
Standard Contractual Clauses for international data transfers
Some data subject request tooling

But you still must:

Ensure your Zaps/scenarios only process data you have legal basis to process
Configure data retention settings within the platform
Document your data flows for GDPR Article 30 records

Neither approach is "automatically compliant." Self-hosting gives you more control at the cost of more work. SaaS shifts infrastructure responsibility to the vendor but adds a third-party processor to your data flow.

What about the US CLOUD Act?

Zapier is a US company. Even with EU data residency, US law enforcement can compel US companies to hand over data stored abroad under the CLOUD Act. This is a real concern for European businesses handling sensitive personal data.

Make is owned by Celonis, a company headquartered in Munich with a secondary office in New York. Make's DPA references the EU-US Data Privacy Framework. n8n GmbH is a German company based in Berlin, and n8n Cloud data stays on Azure in Frankfurt.

Self-hosted n8n on a European VPS owned by a European company (like Virtua.Cloud, a French provider) avoids the CLOUD Act question entirely. No US entity is involved in the data flow.

This matters most for businesses in healthcare, legal services, finance, or any sector where data sovereignty is a regulatory requirement, not just a preference.

Secure n8n with Nginx Reverse Proxy, TLS, and Security Headers

What are the honest tradeoffs of self-hosting vs SaaS automation?

No platform wins everywhere.

Where Zapier wins

7,000+ integrations. The largest app library. If you need a niche SaaS connector, Zapier probably has it.
Zero maintenance. No servers, no updates, no database backups.
Zapier Agents. AI-powered autonomous agents that work across 7,000+ apps. Useful for lead processing, support tickets, and multi-step research tasks.
Fastest to start. Sign up and build a Zap in minutes.

Where Make wins

Visual debugging. The execution replay shows data flowing through each module. The best debugger of the three for complex scenarios.
Credit pricing. Generally cheaper than Zapier at comparable scales because credits cost less than tasks.
EU hosting on all plans. Select EU at signup. No enterprise plan required for EU data residency.
AI Agents with reasoning panels. Visual transparency into how agents make decisions. Module Tools let you turn any module into an AI-callable tool.

Where n8n wins

Unlimited executions (self-hosted). Pay only for infrastructure.
Full data control. Everything stays on your server.
Code when you need it. JavaScript/Python nodes for custom logic that no-code platforms cannot handle.
LangChain integration. Built-in AI agent nodes using the LangChain framework. Connect any LLM, add memory, tools, and output parsers directly in the workflow editor.
400+ integrations and growing. Community-contributed nodes fill many gaps.

Where each platform falls short

Pain point	n8n	Zapier	Make
Server maintenance	You handle updates, backups, uptime	N/A	N/A
Integration breadth	400+ (vs 7,000+ Zapier)	Best in class	2,000+
Learning curve	Steeper for non-technical users	Easiest to start	Moderate
Cost at scale	Cheapest	Most expensive	Middle ground
AI capabilities	LangChain nodes (flexible but technical)	Agents + Copilot (easiest AI)	AI Agents + reasoning panels
License	Fair-code (not OSS)	Proprietary SaaS	Proprietary SaaS

A note on n8n's license: many comparison articles call n8n "open source." It is not. Since 2022, n8n uses the Sustainable Use License, a fair-code license. You can view and modify the source code. You can run it for internal business use. You cannot redistribute it commercially. This matters if you plan to build a product on top of n8n.

How do the AI automation features compare?

All three platforms added AI capabilities in 2025-2026. The approaches differ significantly.

n8n integrates LangChain directly into its workflow editor. You pick your LLM (OpenAI, Anthropic, Ollama, any LangChain-compatible model), add memory nodes, tool nodes, and output parsers. This gives maximum flexibility but requires understanding LangChain concepts. On self-hosted n8n, your AI prompts and responses stay on your infrastructure.

Zapier offers Agents and Copilot. Agents are autonomous AI assistants that work across Zapier's 7,000+ integrations. Copilot lets you build workflows using natural language. AI Actions expose Zapier actions to external AI platforms (GPTs, LangChain, LlamaIndex). Easiest to use, but your data flows through Zapier's AI infrastructure.

Make introduced next-generation AI Agents with visual reasoning panels. You can see how the agent thinks, which tools it calls, and why. Module Tools turn any Make module into an AI-callable tool. Strong on transparency and debugging. Multimodal inputs (documents, images, audio) are on the roadmap.

AI Feature	n8n	Zapier	Make
AI agent builder	LangChain nodes	Zapier Agents	Make AI Agents
LLM flexibility	Any LangChain-compatible	Zapier's AI models	Make's AI models + custom providers
Self-host AI data	Yes (self-hosted)	No	No
Natural language builder	AI Builder (experimental)	Copilot	Chat-based agent building
Third-party AI integration	Native (LangChain)	AI Actions API	Module Tools
Visual debugging	Standard execution view	Basic	Reasoning panels (best)

If privacy of AI prompts and responses matters to you, self-hosted n8n is the only option that keeps everything on your infrastructure.

Should you switch from Zapier to n8n?

Migration comes up a lot in automation communities. Here is the practical answer.

Switch if: your monthly Zapier bill is painful, your workflows are complex (5+ steps), your team has someone comfortable with Linux server management, or you have data residency requirements that Zapier's enterprise plan does not satisfy.

Stay on Zapier if: you use fewer tasks than your current plan allows, you rely on niche integrations that n8n does not support, or nobody on your team can maintain a server.

Migration timeline: Most teams report 2-4 weeks to rebuild their highest-volume Zaps in n8n. Start with the workflows that burn the most tasks. n8n's HTTP Request node can replicate most Zapier integrations that do not have a native n8n node.

Migration risk: n8n's node library is smaller. Before committing, check n8n's integration list for every app your workflows use. Missing integrations mean building custom HTTP Request nodes or writing code nodes.

Which automation platform should you choose?

Use this decision matrix to match your priorities.

Priority	Best choice	Why
Lowest cost at scale	n8n self-hosted	12-48 EUR/mo for unlimited executions (VPS cost only)
Lowest cost, low volume	Make Free or Zapier Free	Free tiers available on both
Maximum integrations	Zapier	7,000+ apps, most niche connectors
EU data residency (easy)	n8n self-hosted on EU VPS	Full control, no third-party processor
EU data residency (SaaS)	Make	EU hosting available on all plans
GDPR with least effort	Make or n8n Cloud	EU-hosted SaaS with DPA included
Best visual debugging	Make	Execution replay, reasoning panels for AI
AI flexibility	n8n	LangChain integration, any LLM, self-hosted AI data
Fastest setup, no servers	Zapier	Sign up and automate in minutes
Technical team, custom logic	n8n	JavaScript/Python code nodes, full API access
Non-technical team	Zapier	Lowest learning curve

The short version

Choose n8n self-hosted if you want cost control, data privacy, and you are comfortable managing a Linux server. Pair it with a European VPS for full GDPR data residency.
Choose Zapier if you need the widest integration library, your team is non-technical, and per-task costs at your volume are acceptable.
Choose Make if you want a middle ground: generally cheaper than Zapier, EU hosting available on all plans, and a strong visual workflow builder.
Choose n8n Cloud if you want n8n's execution-based pricing without managing servers. EU-hosted, but execution limits apply.

What about other alternatives?

This comparison focused on the three most popular platforms. Other tools worth evaluating:

Activepieces is an open-source (MIT license) alternative to n8n. Truly open source, but a smaller community and fewer integrations.
Huginn is a self-hosted agent system for data monitoring. More focused on data scraping and notifications than general workflow automation.
Temporal and Prefect are developer-oriented workflow orchestration tools. Better suited for data pipelines than business process automation.

None of these have the integration breadth or community size of n8n, Zapier, or Make. But if n8n's fair-code license is a dealbreaker, Activepieces is worth a look.