Automation March 10, 2026 · 10 MIN

Self-Hosted n8n vs n8n Cloud: Cost, Control, and When Each Wins

n8n Cloud vs self-hosted: a practical decision guide for cost, control, compliance, ops burden, and when each option is the right production choice.

Automation

Self-Hosted n8n vs n8n Cloud: Cost, Control, and When Each Wins

The hosting question for n8n is not ideological. It is whether you want to buy convenience or operate infrastructure — and which constraints (data, cost, network) dominate.

Spurlock Studios runs both depending on the client. This is the decision framework we use. Broader production rules live in the Production n8n handbook.

The short answer

Choose n8n Cloud when you want less ops, faster start, and your compliance posture allows a reputable SaaS workflow host.
Choose self-hosted when you need network placement, data residency, custom proxying, or cost advantages at high volume and someone will actually own the box.
Do not choose self-hosted because it feels more “pro” if nobody will patch it.

Comparison that matters in production

Factor	n8n Cloud	Self-hosted
Time to first production workflow	Faster	Slower (DNS, TLS, backups, upgrades)
Patching & uptime ops	Vendor	You (or your platform team)
Data residency / VPC	Limited to vendor options	Full control
Outbound IP stability / allowlists	Vendor IPs	Your IPs
Cost at low volume	Usually predictable and fine	Often more expensive once you count labor
Cost at high volume	Can climb with plan tiers	Infra may win if labor is already paid
Credential locality	In vendor systems	In your environment
Needs a named ops owner	Light	Hard requirement

Is n8n Cloud worth it?

Yes — for many teams — if:

You are early and need reliability without hiring platform time
Your workflows are standard SaaS-to-SaaS with normal secrets posture
Your buying team accepts vendor subprocessors
You would under-invest in backups and upgrades if you self-hosted

Cloud is not “less production.” Production is discipline: idempotency, DLQ, schemas, approvals. Hosting is just where the rail lives.

Cloud is the wrong default if legal already said “automation payloads cannot leave our network” or if you need private connectors to systems that are not internet-reachable without a bridge you control.

When self-hosted wins

Self-host when two or more are true:

Compliance or customer contracts require it
You must place n8n on a private network next to internal APIs
Execution volume makes Cloud pricing uncomfortable and you have ops capacity
You need custom reverse proxy, mTLS, or egress control beyond Cloud knobs

Minimum self-hosted bar:

Automated backups of the n8n database
Documented upgrade path
TLS
Secrets via env / secret manager
Monitoring and disk alerts
A human who gets paged when it dies

If that list feels heavy, buy Cloud.

Cost shape (without fake precision)

Do not compare only subscription stickers.

Cloud TCO: subscription + workflow design time + vendor limits you work around.
Self-hosted TCO: compute + storage + backups + observability + upgrade labor + idle knowledge (“only Jamie knows how it runs”).

A $0 VPS with an abandoned n8n instance is not cheap. It is a future incident.

Revisit cost quarterly against real execution counts. Teams grow into different answers.

Hybrid patterns

Common and sane:

Cloud for marketing/ops experiments
Self-hosted for finance-adjacent or internal-only systems
Export/import workflows with environment-specific credentials

Avoid: two unmanaged environments with the same name and nobody sure which is live. Label aggressively.

Migration notes

Cloud → self-hosted or the reverse:

Export workflows, map credentials fresh (never copy secrets through chat)
Rebuild webhook URLs and update providers
Replay-test idempotency and error workflows in the new place
Run dual-write or shadow for a short window on critical paths
Keep a rollback: DNS and provider webhook endpoints ready to flip back

Treat migration as a production change, not a weekend hobby.

How this fits tool choice

Hosting is independent from “why n8n.” If you are still choosing rails, see n8n vs Make vs Zapier. If you already know n8n is the rail, pick hosting from constraints above.

Ops checklist for self-hosted (minimum viable)

If any box is unchecked, stay on Cloud or hire platform help before migrating.

“Docker run on a VPS” is a science fair project until this list exists.

Performance and scaling notes

Self-hosted control includes scaling knobs: worker processes, queue mode, DB size, binary data handling. Cloud abstracts some of this.

If you run heavy binary transforms or huge payloads, test early. Webhook bursts need reverse proxy timeouts aligned with n8n and provider retry behavior. Misaligned timeouts create duplicate deliveries — which is why idempotency remains mandatory on both hosting modes.

Compliance conversations

When legal asks “where does data live?”:

Cloud: answer with vendor region options and subprocessors list
Self-hosted: answer with your cloud region, encryption, access controls, and backup locations

Also answer what is in the executions: PII retention in logs can matter more than which brand hosts the container. Pair hosting choice with retention policy from the handbook.

Team skills required

Skill	Cloud	Self-hosted
Workflow design	Required	Required
Production spine	Required	Required
Linux/containers	Optional	Required
DB backup/restore	Optional	Required
Networking / TLS	Optional	Required

Do not assign self-hosted ownership to a marketer who happens to be brave. Bravery is not a restore strategy.

Decision worksheet

Copy/paste for your next architecture review:

Data residency requirement? (yes/no/specify)
Private network dependencies?
Peak executions/month (current and 3×)
Platform hours available per month
Incident response owner
Budget preference: subscription vs labor
Timeline to first production workflow

If (1) or (2) is yes → bias self-hosted.
If (4) is near zero → bias Cloud.
If (3) is huge and (4) is solid → model both costs.
If timeline is this week → Cloud, revisit later.

Backup restore drill

If you self-host and have never restored a backup into a scratch environment, you do not have backups — you have files. Schedule a quarterly restore drill:

Spin ephemeral instance
Restore DB
Confirm workflows load
Confirm credentials re-inject process works
Tear down

Record time-to-restore. That number belongs in your incident plan.

Upgrades without drama

Pin n8n versions. Read release notes for breaking changes. Upgrade staging first. Run a short regression on critical workflows (webhook verify, idempotency, one happy path). Then production.

Avoid auto-latest tags on production containers. Surprise major versions pair poorly with Monday morning lead flow.

Networking patterns

Common self-hosted layouts:

Public webhook receiver + private worker
VPN access to admin UI; only webhook path public
Egress via fixed NAT IPs for vendor allowlists

Cloud cannot always match fixed egress needs. That single requirement has decided self-hosted for several of our clients.

When Cloud becomes worth revisiting

Teams sometimes outgrow their own ops capacity. Signals to move back to Cloud (or to managed hosting help):

Repeated missed upgrades
Single owner burnout
Restore drills skipped twice
Security questionnaire answers getting shaky

Pride is not an availability strategy. Choose the mode you can operate.

Spurlock Studios default bias

Default recommendation: n8n Cloud for speed and clean ops, unless residency, network, or scale economics say otherwise. We implement the same production spine on either. Hosting is a constraint fit — not a personality test.

Closing operating notes

Hosting bravado fades the first time nobody can restore last night’s backup.

Field note from production

The pattern above is not theoretical. When it is missing, the failure mode is predictable: a duplicate side effect, a muted channel, a CRM row that cannot be trusted, or a finance fire drill. When it is present, the workflow becomes boring — which is the goal.

If you only have time for one improvement this week, implement the control this post centers on, wire an owner, and test the failure case once in staging. That single loop does more than another connector.

For the full spine across idempotency, DLQ, schema, approvals, and hosting, keep the Production n8n handbook open while you build. When you want a production review instead of another internal debate, use the automation lane or book a call.

Write the happy path on one page.
Mark irreversible steps.
Add the control from this article before expanding scope.
Prove one failure case in staging.
Ship behind the tightest autonomy setting you can tolerate.
Review metrics in two weeks; only then loosen.

Skipping straight to step 6 is how demos become incidents. Order is part of ROI.

Thirty-day revisit

Revisit hosting thirty days after go-live with three numbers: execution volume, hours spent on platform ops, and incident count. Let those numbers — not aesthetics — decide whether to stay, migrate, or buy help.

FAQ

n8n Cloud vs self-hosted — which is better?

Neither universally. Cloud wins on ops convenience and speed. Self-hosted wins on control, residency, and certain cost/network cases. Pick from constraints and ownership capacity.

Is n8n Cloud worth it for a small team?

Usually yes, if SaaS subprocessors are acceptable. Small teams rarely win by running their own automation platform on nights and weekends.

Is self-hosted more secure?

Not automatically. A patched, monitored Cloud can beat a neglected VM. Self-hosted can be more secure when you need private networking and you operate it well.

Can I start on Cloud and move later?

Yes. Many teams do. Design workflows with portable patterns (explicit credentials, documented webhooks, minimal hard-coded URLs) so migration is boring.

What about execution limits?

Cloud plans have limits and pricing tiers; self-hosted limits are your hardware and database. Model against peak month, not average quiet week.

Who should own self-hosted n8n?

A platform/devops owner with a backup human. “The freelancer who set it up” is not an ownership model.

CTA

Buy convenience until constraints force control — then earn that control with real ops.

If you want a recommendation for your volume and compliance posture, read the handbook, then use automation or book a call.