Your tools, finally talking to each other
n8n is the backbone I use to connect systems that were never designed to work together — self-hosted, documented, and version-controlled.
Hosted platform or self-hosted?
The honest comparison, including the part where self-hosting costs you maintenance time. For small, non-sensitive workloads a hosted platform is still the right answer.
| Consideration | Hosted platform | Self-hosted n8n |
|---|---|---|
| Per-task platform fees | Grows with your volume | Fixed server cost |
| Data location | Vendor infrastructure | Your own perimeter |
| Long-running jobs | Timeout limits apply | Queue mode, no ceiling |
| Custom code | Limited | Full JavaScript and Python nodes |
| Maintenance | Handled by vendor | Yours, or mine on retainer |
Inside the build
Self-hosted means your data stays on infrastructure you control and there is no per-task bill that grows with your success. Documented means the workflows do not live in one person's head. Both matter more than the automation itself once you are past the first month.
- Self-hosted and private — Runs on your server. No per-task pricing, no customer data leaving your perimeter.
- Anything can connect — If it has an API, a webhook or a database, it joins the workflow.
- Failures do not stay silent — Retries, error branches and alerting are built in from the first workflow.
- Handover ready — Naming conventions, diagrams and runbooks so your team can maintain it without me.
What is included
Everything here is delivered, documented and handed over. Nothing is left depending on my availability.
Self-hosted deployment
n8n on Docker with backups, TLS, and monitoring wired in from day one.
Error workflows
Retry with backoff, dead-letter queue, and an alert that reaches a human within a minute.
Credential vaulting
Secrets held properly, with separate development and production environments.
Reusable sub-workflows
Shared components instead of the same nine nodes copied across twelve flows.
Queue mode at volume
Worker processes and Redis for high-throughput jobs that would time out elsewhere.
Written runbook
Every production workflow ships with a diagram and a page explaining what it does and what to do when it fails.
What companies describe, and what gets built
Left is the symptom as it is usually reported. Right is the actual fix.
Zapier costs more every month and still times out on large jobs.
Self-hosted n8n in queue mode handles the same volume at a fixed server cost, with no per-task ceiling.
A workflow broke last Tuesday and nobody noticed until Friday.
Error workflows alert the owner within a minute and hold the failed payload for replay once the cause is fixed.
Only one person understands how the automation works.
Every workflow gets a diagram, a naming standard and a written runbook, plus a handover session.
The path one item takes
Every stage is logged, and every failure has a defined destination.
Where this lands hardest
Technologies used
Measured before and after, by the client
Placeholder figures. Replace with your own verified numbers before launch.
Frequently asked
About n8n Workflow Automation specifically.
Self-hosted in most cases — cheaper at volume and it keeps customer data inside your own environment. Cloud is perfectly reasonable for small, non-sensitive workloads and I will say so when it fits.
It retries on a schedule, then routes the payload to an error queue and alerts the owner. Nothing is silently dropped, and failed runs can be replayed once the cause is fixed.
Yes. That is the point of the naming conventions and the runbook. Handover includes a working session with whoever will own it.
Stored in n8n's encrypted credential store with environment separation, and never written into node parameters or exported JSON.
Related services
Rarely bought alone.
Tell me what is taking too long
Thirty minutes, no pitch. If automation is the wrong answer for your situation, you will hear that on the call.
Reply within one business day · NDA before any system detail is discussed