Every month, a manager asks me "should we go with Zapier? Or Make? Or this thing I heard about, n8n?"

These are three really different tools, often confused because they all solve the same surface problem: connecting tools that don't talk to each other. Here's what I do with them on my engagements, with the figures and the cases where each one wins.

The 3 tools in one sentence

ToolPositioning
ZapierThe standard automation SaaS. Simple, expensive, reliable.
Make (formerly Integromat)The more powerful and more affordable cloud alternative.
n8nOpen source, self-hostable or cloud, the most powerful and the most owned.

Zapier was the pioneer (2011). Make (formerly Integromat, 2012) is the European outsider that gained ground on flexibility and price. n8n (2019) arrived as open source and exploded between 2023 and 2026.

Direct comparison table

Here's my comparison grid after about thirty automation engagements between 2023 and 2026:

CriterionZapierMaken8n
Entry pricing€20/month (750 tasks)€9/month (10,000 ops)Free (self-hosted) or €20/month (cloud)
Pricing at volume (100k+ ops/month)~€400/month~€90/month~€30/month (server) or ~€60/month (cloud)
Ready connectors7000+2000+500+ native, infinite via HTTP
Learning curve1h3-4h1-2 days
Conditional logicBasicExcellentExcellent + code
Data ownershipAt Zapier (USA)At Make (EU possible)At yours, wherever you want
Custom code❌ limited⚠️ basic JS✅ full JS + Python
Strict GDPR compliance⚠️ contractual✅ OK✅ total (self-hosted)
SMB-side maintenanceNoneNoneMedium (if self-hosted)

Zapier — when I recommend it

The rule: maximum simplicity, budget not a concern, low volume.

The ideal case

A solo consultant who wants:

  • New lead in Calendly → added to Notion + Slack notification + Mailjet welcome email
  • Stripe payment → entry in Google Sheet + Pennylane invoice + personal message

5 automations, 500-1500 tasks per month. Zapier at €20/month is the best choice: 30 minutes to configure, runs by itself, zero maintenance.

When to avoid Zapier

  • As soon as volume exceeds 10,000 tasks/month — pricing explodes punitively (Zapier bills per task, not per operation).
  • As soon as the logic gets complex (loops, multiple branches, data transformations) — you're going to suffer.
  • As soon as you need to protect sensitive data — Zapier hosts in the USA and logs all data passing through.

Make — my default choice for most SMBs

The rule: unbeatable price/quality ratio for 80% of SMB cases.

Make is where I go first when a client tells me "we want to automate". Three reasons:

1. The pricing model is healthy

Make bills by operation (each step in a scenario), not by triggered task. Concrete result:

Monthly volumeZapierMake
1,000 ops/tasks€20€9
10,000€50€9
100,000€400€29
1,000,000~€1,500€79

For an SMB ramping up volume, the gap becomes abyssal.

2. Visual logic is better

The Make interface is a real visual canvas where you see flows, branches, aggregations. It's more readable than Zapier's linear queue for any non-trivial automation.

3. EU hosting available

Since 2024 Make offers Europe-hosted instances. GDPR-friendly without negotiation.

When to avoid Make

  • Massive volumes and tight budget — beyond 500k ops/month, self-hosted n8n becomes unbeatable.
  • Need for Python code or specific libraries — Make only does basic JS.
  • Highly sensitive data (health, regulated finance) where absolute sovereignty matters — switch to self-hosted n8n.

n8n — my choice for power and ownership

The rule: you want ownership, you have volume, or you have hard compliance requirements.

n8n arrived in 2019, and between 2023 and 2026 it grabbed huge market share. Why:

1. Open source, really

You can self-host for free on a small server (€15-30/month at Hetzner or OVH). Your workflows no longer depend on a US SaaS vendor.

2. Native power-user

  • JavaScript and Python code in nodes — you can do anything.
  • Native HTTP Request, ultra-flexible — connects any API even without a dedicated connector.
  • Rich expressions and templates — clean data manipulation.

3. Native AI integration

Since 2024, n8n natively integrates:

  • Claude, GPT, Gemini, Mistral nodes
  • Vector nodes (Supabase Vector, Pinecone, Qdrant)
  • AI agents with memory and tools

It's literally the most advanced platform to build custom AI agents without restarting from scratch in code.

When to avoid n8n

  • No technical appetite or no one in-house — self-hosted n8n requires a minimum of know-how. The cloud version exists (€20/month) but loses the ownership advantage.
  • You just want 3 small automations — Make is faster to deploy, less setup.

Comparison on a concrete case: automating invoicing

Imagine the workflow: Stripe invoice paid → entry into Pennylane accounting + personal email to client + Slack notification + PDF archive in Google Drive.

Volume: 500 invoices/month = 500 Zapier tasks = ~2,500 Make / n8n operations.

ToolSetup timeMonthly costMaintenance
Zapier45 min€20 (Starter plan)None
Make1h€9 (Core plan)None
n8n cloud1h30€20None
n8n self-hosted2h (+30 min server setup)€15 (VPS server)30 min/month

At 500 invoices, costs are close. But the same configuration at 5,000 invoices/month:

ToolMonthly cost
Zapier~€120
Make€16
n8n cloud€50
n8n self-hosted€15

The gap widens fast.

My final decision grid

Three questions are enough:

Question 1 — Expected monthly volume

  • Less than 1,000 ops/month → Zapier or Make (fast, cheap)
  • Between 1,000 and 100,000 ops/monthMake (unbeatable sweet spot)
  • More than 100,000 ops/monthn8n (self-hosted or cloud)

Question 2 — Data sensitivity

  • Public or low-sensitivity data → any of the three
  • Standard client data under GDPR → Make or n8n
  • Highly sensitive data (health, regulated finance, industrial secrets)self-hosted n8n only

Question 3 — Do you want ownership?

  • No, I just want it to run → Zapier or Make
  • Yes, I want code and data to stay with meself-hosted n8n

The truth I keep insisting on

None of these three tools is the best in absolute terms. They are optimal in different contexts.

What is always a mistake:

  • Picking Zapier "because we know it" without looking at the 18-month cost.
  • Picking n8n "because it's open source" without anyone to maintain it.
  • Stacking all 3 because each team picked their own — you end up with €150/month of tools doing the same thing.

When none of the 3 is enough

There are cases where I recommend going custom code (Node, Python) from the start:

  • Very specific business logic that has no place in no-code.
  • Need for extreme performance (latency < 100ms on 10k req/sec).
  • Dependency on internal proprietary libraries.
  • Massive volume (millions of operations per day) where no-code overhead costs more than custom dev.

In these cases, n8n or Make become orchestrators calling custom services I've coded alongside.

Where to start

If you want me to look at what fits your case, 30 minutes are enough. I'll tell you honestly if Zapier is OK for you, if you should switch to Make or invest in n8n — and above all if your project really needs one of these tools or something else.

To see my price ranges on automation engagements, it's here.


To go further