n8n Self-Hosted vs Make.com Enterprise: The Real Cost Breakdown at 1M Tasks/Month

Porfespejelarroyo@gmail.com

n8n Self-Hosted vs Make.com Enterprise: Real Cost at 1M Tasks/Month

Choosing an automation backbone at enterprise scale usually comes down to a fundamental architectural dilemma: pay for operational convenience or take control of your compute infrastructure.

When your organization hits 1,000,000 tasks per month, the pricing disparity between managed iPaaS solutions and self-hosted workflow engines stops being a minor line item and turns into a board-level budget discussion.

Executive Verdict: Make.com wins on zero-maintenance reliability, native app connectors, and immediate time-to-market. However, at 1,000,000 operations per month, n8n Self-Hosted reduces direct software/compute spend by over 88%, even when accounting for managed databases, Redis queuing, and standard DevOps maintenance hours.

The Math at Scale: 1,000,000 Operations Analyzed

Before comparing dollar amounts, we must address the fundamental architectural difference in how both platforms bill for execution.

  • Make.com bills by Operation: Every single module execution within a scenario burns exactly 1 operation. A webhook triggering a router with 4 filter branches and 3 API calls can easily consume 8 to 12 operations per single inbound event.
  • n8n bills by Workflow Execution: In self-hosted n8n, there are no artificial operation limits. A single execution processing an array of 5,000 records through 10 transformation nodes consumes compute resources (CPU/RAM), not licensing credits.
MetricMake.com (Enterprise/Scale Tier)n8n Self-Hosted (Queue Mode)
Billing UnitOperations / Module runsPure Cloud Compute (vCPU/RAM)
Direct Monthly Cost (1M units)~$800 – $1,250 / mo~$65 – $145 / mo
Payload Size LimitsStrict limits per plan (5 MB – 50 MB)Limited only by host RAM
Execution TimeoutMax 40–60 minutesUnlimited (configurable via Node.js)
Data ResidencyShared EU/US Cloud regions100% on-premise / Private VPC

Make.com Pricing Reality: The Overage Trap

Make’s tiered model is predictable at low volumes (10,000 to 100,000 operations). But at enterprise throughput, three structural friction points emerge:

  1. The Multi-Module Multiplier: An event-driven architecture ingesting 100,000 webhooks per month rarely uses 100,000 operations. With transformations, routers, and external API error-handlers, that volume turns into 800,000 to 1,500,000 operations.
  2. Overage Spikes: If a rogue script or third-party retry storm floods your webhooks over a weekend, auto-purchased operation tiers kick in immediately, leading to unanticipated monthly invoice spikes.
  3. Data Security Overhead: If you handle HIPAA, PCI-DSS, or strict GDPR PII, Make’s standard shared-cloud infrastructure may force an upgrade to dedicated Enterprise contracts starting in five-figure annual commitments.

n8n Self-Hosted Infrastructure Blueprint (Production-Ready)

To process 1,000,000 executions per month without memory leaks or execution bottlenecks, you cannot run n8n in basic default single-process mode. You need n8n Queue Mode powered by Redis and an external PostgreSQL database.

Incoming Webhooks / APIs Payload Ingestion Reverse Proxy (Traefik / Nginx) SSL Termination & Routing n8n Main Instance Editor UI & Orchestrator PostgreSQL 16 State & Execution Persistence Redis 7 In-Memory Job Broker (BullQueue) n8n Worker 01 Dedicated Node Execution n8n Worker 02 Dedicated Node Execution

Production Docker Compose Stack

version: '3.8'

services:
  postgres:
    image: postgres:16-alpine
    restart: always
    environment:
      POSTGRES_USER: n8n_admin
      POSTGRES_PASSWORD: ${DB_PASSWORD}
      POSTGRES_DB: n8n_data
    volumes:
      - db_storage:/var/lib/postgresql/data

  redis:
    image: redis:7-alpine
    restart: always
    command: ["redis-server", "--appendonly", "yes"]
    volumes:
      - redis_storage:/data

  n8n-main:
    image: docker.n8n.io/n8nio/n8n:latest
    restart: always
    environment:
      - DB_TYPE=postgresdb
      - DB_POSTGRESDB_HOST=postgres
      - DB_POSTGRESDB_USER=n8n_admin
      - DB_POSTGRESDB_PASSWORD=${DB_PASSWORD}
      - DB_POSTGRESDB_DATABASE=n8n_data
      - EXECUTIONS_MODE=queue
      - QUEUE_BULL_REDIS_HOST=redis
      - N8N_ENCRYPTION_KEY=${ENCRYPTION_KEY}
    ports:
      - "5678:5678"
    depends_on:
      - postgres
      - redis

  n8n-worker:
    image: docker.n8n.io/n8nio/n8n:latest
    restart: always
    command: /usr/local/bin/n8n worker
    environment:
      - DB_TYPE=postgresdb
      - DB_POSTGRESDB_HOST=postgres
      - DB_POSTGRESDB_USER=n8n_admin
      - DB_POSTGRESDB_PASSWORD=${DB_PASSWORD}
      - DB_POSTGRESDB_DATABASE=n8n_data
      - EXECUTIONS_MODE=queue
      - QUEUE_BULL_REDIS_HOST=redis
      - N8N_ENCRYPTION_KEY=${ENCRYPTION_KEY}
    depends_on:
      - postgres
      - redis

volumes:
  db_storage:
  redis_storage:

Direct Infrastructure Cost Breakdown

  • Compute (Hetzner Cloud / AWS EC2):
    • 1x CPX31 (4 vCPU, 8 GB RAM) for Main + Postgres + Redis: $16.50/mo
    • 2x CPX21 (3 vCPU, 4 GB RAM) for Execution Workers: $18.00/mo
  • Block Storage & Offsite Backups (S3 / R2): $5.00/mo
  • Managed Network & Ingress (Cloudflare Pro / Traefik): $20.00/mo
  • Total Direct Hardware Cost: ~$59.50 / month

The Hidden Overhead: Engineering Hours & TCO

Hosting your own infrastructure introduces operational friction that pure SaaS pricing avoids. To calculate true Total Cost of Ownership (TCO), consider human engineering time:

  • Setup & Production Hardening: ~8 hours upfront (One-time cost: ~$800).
  • Monthly Patching & Node Upgrades: ~1.5 hours/month (~$150/month).
  • Database Pruning: Storing 1,000,000 workflow executions without automated cleanup scripts will crash PostgreSQL within 90 days. You must configure execution pruning (EXECUTIONS_DATA_PRUNE=true).

The 1-Year TCO Comparison (1M Operations/Month)

Cost CategoryMake.com Enterprisen8n Self-Hosted
Annual Software/Compute$11,880$714
Annual Maintenance (DevOps @ $100/hr)$0$1,800
Initial Deployment Setup$200$800
Total Year 1 Investment$12,080$3,314
Total Year 2+ Run-rate$11,880 / year$2,514 / year

Architectural Comparison Matrix

Technical CapabilityMake.comn8n Self-Hosted
Custom Code ExecutionBasic JS (Strict Sandbox)Full Node.js, Python, npm modules
Version Control & CI/CDEnterprise only / manual JSONNative Git Integration via CLI/API
Debugging Heavy PayloadsTruncates large arrays in UIFull variable inspection in browser
Community Custom NodesClosed ecosystemOpen npm ecosystem / Custom Docker builds
Local LLM / Private AICloud APIs only (OpenAI, Anthropic)Direct connectivity to self-hosted Ollama / vLLM

Final Decision Framework: Which One Should You Pick?

Strictly regulated data or local LLMs? HIPAA, Banking PII, On-premise AI models YES Choose n8n Self-Hosted Air-gapped / Private VPC security NO In-house DevOps & Linux skills? Ability to manage Docker, Postgres & Redis NO Choose Make.com Zero maintenance justifies SaaS premium YES Volume above 500k monthly tasks? Scale where SaaS overage tiers multiply YES Choose n8n Self-Hosted Saves $8,000+ / year in run-rate NO Choose Make.com Fastest time-to-delivery for small stacks

Choose Make.com if: Your automation team consists primarily of non-technical operators, your integrations rely heavily on standard SaaS tools (Slack, HubSpot, Airtable), and paying a premium is preferable to maintaining server uptime.

Choose n8n Self-Hosted if: You process repetitive bulk data, need custom npm/Python logic, require strict regulatory compliance, or want to deploy local AI pipelines without variable per-token automation penalties.

Aquí tienes la sección de Preguntas Frecuentes (FAQ) redactada en inglés técnico, lista para copiar y pegar justo antes de la conclusión o llamada a la acción de tu artículo.

Además de responder a las dudas de intención de búsqueda transaccional, coincide con las preguntas incluidas en el marcado Schema JSON-LD:

Frequently Asked Questions

Is self-hosting n8n really cheaper than Make.com at scale?

Yes. At 1,000,000 monthly operations, Make.com plans cost between $800 and $1,250/month depending on your operation tier and overage fees. A resilient, production-ready n8n setup deployed on cloud compute (such as Hetzner or AWS EC2) costs roughly $60 to $145/month in compute and storage. Even when budgeting 1.5 to 2 hours of DevOps maintenance per month, self-hosted n8n reduces total operational software spend by more than 80%.

What infrastructure do I need to run n8n at 1,000,000 executions per month?

Running n8n at this volume requires Queue Mode rather than the single-process default container. The recommended architecture includes:

  • Orchestrator (Main Instance): 4 vCPU, 8 GB RAM.
  • Database: Managed PostgreSQL 16 (configured with automated execution pruning).
  • Broker: Redis 7 (handling task queuing and concurrency).
  • Workers: At least 2 worker nodes (2–4 vCPU each) to process jobs asynchronously without blocking the UI.

Does n8n require coding knowledge compared to Make.com?

For standard API integrations and basic webhooks, n8n offers a visual drag-and-drop node experience similar to Make. However, managing a self-hosted production cluster requires foundational DevOps skills (Docker, reverse proxies, environment variables, and PostgreSQL maintenance). If your team lacks Linux server administration experience, Make’s managed platform eliminates that infrastructure overhead.

How does n8n handle execution data pruning?

By default, saving millions of execution logs will rapidly exhaust your database storage. In production, you must set EXECUTIONS_DATA_PRUNE=true and define EXECUTIONS_DATA_MAX_AGE (typically 168 hours / 7 days) in your environment variables. This automatically deletes old execution payloads and keeps PostgreSQL lean and fast.

fespejelarroyo@gmail.com