Self-hosted deliverability automation with n8n

If you run infra in-house, paying a SaaS for "operations" to chain HTTP calls feels wrong. n8n flips that: self-host it, wire it to Inbox Check, and you get Zapier- style flows without per-run fees. This guide gets you from a blank VM to a scheduled deliverability workflow in about 30 minutes.

Step 1 — Run n8n via docker-compose

Pick a VM with 2GB RAM. The compose file below runs n8n behind a reverse proxy with Postgres for persistence.

# docker-compose.yml
services:
  n8n:
    image: n8nio/n8n:latest
    restart: unless-stopped
    ports: ['127.0.0.1:5678:5678']
    environment:
      - DB_TYPE=postgresdb
      - DB_POSTGRESDB_HOST=db
      - DB_POSTGRESDB_DATABASE=n8n
      - DB_POSTGRESDB_USER=n8n
      - DB_POSTGRESDB_PASSWORD=${POSTGRES_PASS}
      - N8N_HOST=n8n.yourdomain.com
      - N8N_PROTOCOL=https
      - WEBHOOK_URL=https://n8n.yourdomain.com/
      - GENERIC_TIMEZONE=UTC
      - N8N_ENCRYPTION_KEY=${N8N_ENCRYPTION_KEY}
    volumes:
      - ./data:/home/node/.n8n
    depends_on: [db]
  db:
    image: postgres:16
    restart: unless-stopped
    environment:
      - POSTGRES_DB=n8n
      - POSTGRES_USER=n8n
      - POSTGRES_PASSWORD=${POSTGRES_PASS}
    volumes:
      - ./pg:/var/lib/postgresql/data

Put Caddy or nginx in front for TLS. Generate the N8N_ENCRYPTION_KEY with openssl rand -hex 32.

Step 2 — Store the Inbox Check key

In n8n, go to Credentials → New → Generic HTTP Header Auth. Name Authorization, value Bearer ic_live_xxx. Save. Every HTTP Request node in this workflow references it.

Step 3 — The workflow

Four nodes. Add a Slack node for alerts and a Postgres node if you want history.

1. Schedule Trigger

Every 30 minutes (or cron expression).

2. HTTP Request — create test

Method:     POST
URL:        https://check.live-direct-marketing.online/api/tests
Authentication: Generic Credential / Header Auth (set above)
Body Content Type: JSON
Body:
{
  "sender":   "hello@yourdomain.com",
  "subject":  "Scheduled probe {{ $now.toISO() }}",
  "html":     "<p>Probe content</p>",
  "providers": ["gmail","outlook","yahoo","mailru","yandex"]
}

3. Wait + HTTP Request — poll

Use the Wait node (60 seconds), then an HTTP Request that hits GET /api/tests/{{ $json.testId }}. Wrap in a Loop Over Items with an IF node that re-queues if status is not done, up to 3 attempts.

4. Code node — compute inbox rate

// Code node input = the GET /api/tests/:id response
const providers = items[0].json.providers;
const inbox     = providers.filter(p => p.folder === 'inbox').length;
const rate      = inbox / providers.length;
return [{
  json: {
    ...items[0].json,
    inboxRate: Math.round(rate * 100),
    alert:     rate < 0.80,
    perProvider: providers.map(p => `${p.name}: ${p.folder}`).join(', '),
  },
}];

5. IF → Slack / Postgres

If alert === true, send a message via the Slack node to your deliverability channel. Always log the row to Postgres via the Postgres Insertnode for a long-term trend.

Importing the workflow

We publish the workflow JSON on GitHub. In n8n, click the burger menu → Import from file, pick the JSON, bind credentials to the placeholders, enable the workflow. Done.

Hardening

• Basic auth on the n8n UI via N8N_BASIC_AUTH_ACTIVE=true. Or front it with an OIDC proxy.
• Rotate the encryption key on a schedule; n8n re-encrypts credentials on restart.
• Queue mode if you run more than a few workflows. Redis-backed job queue + separate worker containers.
• Backups. Snapshot the ./data and ./pg volumes nightly. Without these, a lost VM = lost workflows.

Ideas to extend

• Per-ESP monitoring. Duplicate the workflow for each ESP sender. Aggregate into one Postgres table.
• Anomaly detection. Add a Code node that compares today's rate to the rolling 7-day average. Alert only on a >10 point drop.
• DNS + auth health checks. Chain a second HTTP call to the auth endpoint; alert if SPF or DMARC verdict regresses.
• Grafana dashboard. Point Grafana at the Postgres table for a time-series chart of inbox rate.

FAQ