Handoff Day: Why I Won't Ship Make.com to a Paying Client

By Lazar Milicevic · Published July 4, 2026 · 6 min read

A client called me at 11 PM asking why their invoicing bot stopped mid-month. The answer wasn't a bug. It was a $29 paywall inside somebody else's Make.com account. If you build automations for SMB clients and you're still picking between Make.com and n8n on UI feel, you're solving the wrong problem — the one that matters is what happens on handoff day.

The Math That Kills Make.com At Real Client Volume

Here's the actual scenario I run into every week. A small business hires you to automate invoicing. Round numbers: 2,400 invoices/month. Each invoice fires a webhook, pulls customer data from a CRM, generates a PDF, emails it, logs it to Google Sheets, and updates the accounting system. Six operations per invoice on a good day.

Do the math:

2,400 invoices × 6 operations = 14,400 operations/month

Make.com's Core tier at $29/mo ships with 10,000 ops. Your client burns the entire monthly allowance in roughly eight days. Then the automation stops. No error banner. No email. It just stops firing until the next billing cycle — or until someone upgrades.

Next tier is $59/mo (Pro, 10K ops but with more features), and at real SMB scale you land on Teams at $99+/mo. The uncomfortable part nobody says out loud: that subscription lives either in your Make.com account (you're the perpetual middleman) or in the client's (they now need a Teams plan just to own the workflow you built for them). Somebody is bleeding recurring cost for infrastructure that should be commodity.

Compare that to a $6/mo VPS running n8n. 14,400 ops or 14 million ops — the VPS bill is flat.

Per client, per year: $1,116 difference. Multiply across 5 clients, that's over $5,500/year in infrastructure cost you've either saved them or absorbed yourself.

What Actually Happens On Handoff Day

Handoff day is the day you get paid the final invoice and walk away. If you built it right, the client never calls you at 11 PM. If you built it on someone else's subscription, they will.

On self-hosted n8n, handoff is a checklist:

• Export the workflow(s) as JSON from the n8n UI
• Hand over root/SSH access to the VPS (or transfer the account at the VPS provider)
• Rotate the API credentials the client already owns (Stripe, Google, their CRM)
• Deliver a one-page runbook: how to restart n8n, where logs live, how to restore from backup

Total handoff time: under an hour. The client's technically-inclined operations person can maintain it. If they hire an internal dev later, that person inherits a self-contained system running on standard Docker — not a subscription cage.

On Make.com, handoff is one of two bad options:

1. Transfer the account — client now owns a $99/mo subscription they didn't ask for, and if they ever want to move off Make.com, there's no clean export. Scenarios don't serialize to portable JSON the same way n8n workflows do.
2. Keep the account — you're now the permanent middleman. Every credential rotation, every ops overage, every "why did it stop?" goes through you. You didn't ship an automation. You signed up for lifetime support.

That's not a comparison of tools. That's a comparison of business models.

The 10-Minute n8n Install I Actually Use

This is the boring, repeatable version. Any $6/mo VPS from Hetzner, DigitalOcean, or Vultr will run it comfortably for a single client's workload.

# On a fresh Ubuntu 22.04 VPS
apt update && apt install -y docker.io docker-compose

mkdir -p /opt/n8n && cd /opt/n8n
cat > docker-compose.yml <<'EOF'
services:
  n8n:
    image: n8nio/n8n:latest
    restart: always
    ports:
      - "127.0.0.1:5678:5678"
    environment:
      - N8N_HOST=n8n.clientdomain.com
      - N8N_PROTOCOL=https
      - WEBHOOK_URL=https://n8n.clientdomain.com/
      - GENERIC_TIMEZONE=America/New_York
      - N8N_ENCRYPTION_KEY=${N8N_ENCRYPTION_KEY}
    volumes:
      - ./data:/home/node/.n8n
EOF

docker compose up -d

Front it with Caddy for automatic HTTPS:

n8n.clientdomain.com {
    reverse_proxy 127.0.0.1:5678
}

Point a subdomain at the VPS IP, wait for the DNS to propagate, and you're live. From cold VPS to logged-in n8n editor: under 15 minutes on any provider I've used.

The handoff bundle I ship

• docker-compose.yml and Caddyfile in a /opt/n8n folder the client can tar and move
• A daily cron job that dumps ./data to S3-compatible storage (Backblaze B2 is ~$0.005/GB/mo)
• A README.md with restart, backup restore, and credential rotation steps
• Exported JSON of every workflow, stored in a Git repo the client owns

If the VPS dies at 3 AM, restoring the entire system on a new box is a 20-minute exercise for anyone comfortable with SSH.

The One Case Where Make.com Actually Wins

I want to be honest, because "n8n beats Make on everything" is a lie sold by n8n influencers who haven't shipped to a paying client.

Make.com is the correct answer when all three of these are true:

• The client is completely hands-off and will never want to touch the automation
• You're charging a recurring managed-service retainer and absorbing the ops cost inside your margin
• The workflow lives comfortably under 10K ops/month, forever

In that world, Make.com's visual polish is genuinely nicer for demos and the scenario history UI is faster to debug than n8n's execution log. You keep the workflow in your account, you charge a flat monthly, and everyone is happy — as long as you're honest that you're selling a managed service, not an automation. The moment the client asks "can I move this in-house?", you have nothing to hand them.

That's a legitimate business. It just isn't what most SMB clients think they're buying when they hire you.

The Four-Question Framework I Use Before Quoting

I don't debate this on tool feel anymore. I run every prospect through four questions before I send a proposal.

1. Will this workflow exceed 5,000 operations/month within 12 months? If yes → self-hosted n8n. Make.com's op-based pricing punishes growth.
2. Does the client want to eventually own and maintain this themselves? If yes → self-hosted n8n. The JSON export + Docker install is a real deliverable. A Make.com scenario is not.
3. Is this an ongoing managed service where I'm charging recurring and eating platform cost? If yes → Make.com is fine. My margin, my choice.
4. Regulated industry with data residency requirements (HIPAA, financial services, EU customers)? If yes → self-hosted n8n on the client's preferred cloud (or on-prem). No exceptions. Make.com's EU zone helps but doesn't solve customer-controlled key management.

Three out of four questions push you to self-hosted n8n. That matches my client mix — roughly 4 out of 5 projects.

Payback timeline, real numbers

• Setup time to build a solid invoicing workflow: 4–6 hours in either tool
• VPS provisioning + n8n install + HTTPS: ~30 minutes (once you've done it twice)
• Break-even vs Make.com Teams tier: month one
• Handoff on n8n: under an hour
• Handoff on Make.com: a signed managed services agreement

The referral math also matters. Clients who inherit a self-contained system tell their business network you built them something they own. Clients locked into your Make.com account tell their business network they can't leave you. Only one of those generates warm intros.

Why bizflowai.io helps with this

Most of the SMB workflows we deploy at bizflowai.io — invoicing, lead intake, client onboarding, follow-up sequences — sit on self-hosted n8n boxes we hand over on day one, with the JSON exports, credentials, and runbook going to the client's ops person, not staying in our account. When a workflow genuinely fits the managed-service model (low ops, hands-off client, recurring retainer), we'll run it on Make.com and be upfront that the client is buying a service, not owning software. Two different pricing models, two different deliverables, one honest conversation up front.

Want more like this?

I publish practical AI automation, GenAI engineering, and faceless content workflows on YouTube every week.

Subscribe to bizflowai.io on YouTube — never miss a new tutorial.

Planning an AI automation project or need a second opinion on your architecture?

Connect with me on LinkedIn — Lazar Milicevic, GenAI Engineer & bizflowai.io Founder.

Visit bizflowai.io for our services, case studies, and AI consulting.