n8n · 4 workflows · 1,284 executions today
Managed n8n hosting, without the execution counter
Your own n8n instance on UK infrastructure we keep patched, backed up and online. Run as many workflows and executions as your plan's resources allow — we do not meter them.
execution 8f2c1a — Shopify order → Xero → Slack
- Unlimited executions
- UK data centres
- Upgrades handled for you
From £8.99 a month. Prices exclude VAT. No setup fee, no minimum term.
the options
Three ways to run n8n
n8n is free and open source, so the question is never whether you can run it. It is who patches it at 2am and who is responsible when a webhook stops answering.
If you run fewer than a couple of hundred executions a month and enjoy server admin, a £5 VPS is genuinely cheaper. This is for people whose automations have become load-bearing.
metered executions
2,500 executions/month
unlimited executions
| Self-host on a VPS | n8n Cloud | The Email Shop | |
|---|---|---|---|
| Executions | Unlimited | Metered — Starter includes 2,500/month | Unlimited, capped only by your plan's CPU and memory |
| Typical monthly cost | £5–£16 for the server, plus your time | About £17 Starter, about £43 Pro | From £8.99, all in |
| Version upgrades | You, by hand, hoping nothing breaks | Automatic | We upgrade, and roll back if a workflow breaks |
| HTTPS and webhook URL | You configure a reverse proxy and certificates | Included | Included, with a fixed URL that survives restarts |
| Backups | Whatever you remembered to set up | Managed | Nightly database snapshots kept 14 days, restore on request |
| Where your data sits | Wherever you put it | Vendor's regions | UK data centres |
| Community nodes | Yes | Restricted on lower tiers | Yes — install any community node |
What managed actually means here
Not a control panel with n8n bolted on. A dedicated instance, isolated from other customers, that we operate.
No execution metering
Run a workflow every minute if that is what the job needs. Your ceiling is the CPU and memory on your plan, not a counter that resets on the first of the month.
Postgres, not SQLite
Every instance gets a real Postgres database from the start. SQLite is fine until two workflows write at once, and then it is not.
Queue mode available
On Professional and above we can run n8n in queue mode with Redis and separate workers, so one slow HTTP node stops blocking everything behind it.
Locked-down by default
Basic auth or SSO in front of the editor, HTTPS enforced, and the instance is not shared with anyone else's workflows or credentials.
Nightly backups, 14-day history
Database and credentials snapshotted every night. Ask us and we will restore to any of the last fourteen days, usually within the hour.
Support that reads your workflow
Our team runs n8n themselves. If a node has started failing, you will get an answer about that node, not a link to the documentation.
The four things that break a self-hosted n8n
In roughly that order. None of them are difficult on their own; all of them arrive at an inconvenient moment.
The webhook URL changes and nothing tells you
n8n builds callback URLs from WEBHOOK_URL. Move the container, change the port or restart behind a new tunnel and every third-party service you registered is still posting to the old address. Failures are silent, because the caller gets an error and you get nothing. Every instance we run has a fixed hostname and certificate from day one, so the URL you give Stripe or Shopify keeps working.
SQLite hits its concurrency wall
The default database is a single file with a write lock. It is completely fine for one workflow, and it starts throwing 'database is locked' the first time two executions finish at the same moment. The migration to Postgres is straightforward once, and miserable under pressure at the point you actually need it. We start every instance on Postgres so the wall is not there.
An upgrade changes a node's behaviour
n8n ships often, which is a good thing until a minor version alters how a node handles an empty field and a workflow that ran for a year quietly starts producing nothing. We upgrade on a schedule, keep the previous version ready, and roll back if something you depend on changes. On Professional and above you get a staging instance to try the new version first.
One slow HTTP request blocks everything
In the default single-process mode, executions share one Node process. A workflow waiting thirty seconds on a slow API is thirty seconds nothing else runs. Queue mode fixes it by moving executions to separate workers with Redis in between — it is a change to how the whole thing is deployed, which is why it tends not to happen until it hurts. It is available from Professional upward and we configure it for you.
Straightforward monthly pricing
One rate that does not change after the first year. Cancel any month.
Starter
A first production instance
£8.99/month
- 1 vCPU
- 2 GB RAM
- 20 GB NVMe
- Postgres included
- Unlimited workflows and executions
- Automatic version upgrades
- Nightly backups, 14-day history
- HTTPS and a fixed webhook URL
- 24/7 UK support
Professional
Automations the business depends on
£16.99/month
- 2 vCPU
- 4 GB RAM
- 60 GB NVMe
- Queue mode + Redis
- Everything in Starter
- Queue mode with a dedicated worker
- Staging instance for testing upgrades
- Custom domain for your editor
- Priority support queue
Business
High volume, or client work
£29.99/month
- 4 vCPU
- 8 GB RAM
- 120 GB NVMe
- Multiple workers
- Everything in Professional
- Three workers for parallel executions
- Isolated instance per client on request
- Hourly database snapshots
- Named technical contact
Prices exclude VAT. No setup fee, no minimum term.
Frequently asked questions
Move your automations somewhere they will be looked after
Free migration from any existing n8n instance, and a UK team who run the same software. If you would rather talk it through first, we are on the phone.