n8n Cloud vs. Self-Hosted: The Decision You Need to Make Before You Build

Before you build a single workflow, you have to answer one question: where does n8n actually run? This is not a detail you can leave for later. It decides who can see your client data, what you pay every month, and who you call at 11pm when something breaks.

There are two ways to run n8n, the orchestration platform behind all 12 Plays in The AI Workforce Playbook. You can use n8n Cloud, where n8n runs the software for you on a subscription. Or you can self-host, where n8n runs on a server you rent and control. Both run the exact same software. The difference is entirely about who is responsible for the server underneath it.

This guide explains the tradeoff in plain English, gives you real numbers, and ends with a decision framework based on your firm size and whether you have anyone technical on the bench.

Use n8n Cloud if you have no technical staff, you want to start this month, and your client data is not subject to strict residency or confidentiality rules.

Self-host if you handle regulated data (law, finance, healthcare-adjacent), you want full control over where client information lives, or you expect to run high volume and want to avoid per-execution pricing.

Most firms reading the book land on self-hosting, not because it is cheaper at small scale (it barely is), but because the firms that buy this book usually care a lot about where client data sits. If that is you, skip to the data section.

n8n Cloud is the managed version. You sign up, you get a workspace at a URL like yourfirm.app.n8n.cloud, and you start building. n8n handles the server, the updates, the backups, and the uptime. You handle the workflows.

It is a monthly subscription. Pricing is based on how many workflow executions you run and how many active workflows you keep running, not on how many steps each workflow has. As of this writing, the Starter plan runs around 20 to 25 dollars a month for roughly 2,500 executions, and the Pro plan runs around 50 to 65 dollars a month for higher volume and more active workflows.

The appeal is simple: you never touch a server. There is no setup beyond creating an account. If something breaks at the infrastructure level, that is n8n's problem to fix, not yours. For a firm with no IT person, that is worth real money.

The catch is that your workflow data, your execution logs, and anything passing through your automations travel across n8n's infrastructure, which sits on AWS in the EU. For most firms that is perfectly fine. For some, it is a non-starter.

Self-hosting means you rent a server, called a VPS (virtual private server), from a provider like DigitalOcean, Google Cloud, or Microsoft Azure, and you install n8n on it. The software is free and open-source, so you pay nothing for n8n itself. You pay only for the server, which for a firm of 10 to 250 people is usually a 20 to 50 dollar per month VPS.

DigitalOcean offers a one-click n8n app that gets you from zero to a running instance in about 30 to 60 minutes, even if you have never touched a server before. We walk through the providers in the VPS hosting comparison, and there is a full step-by-step in the n8n self-hosting setup guide for DigitalOcean.

The thing self-hosting gives you that Cloud cannot: total control. The server is yours. Your client data lives on infrastructure you own and nobody else has a copy. You decide where in the world it sits. You can run unlimited executions for the flat cost of the server. There is no per-execution meter ticking.

The thing self-hosting costs you: someone has to keep it running. Software updates, security patches, backups, and the occasional reboot are now your job. It is not a heavy lift, but it is a recurring one. Budget two to four hours a month, and decide upfront who owns it.

Cost is the factor most people fixate on, and it is usually the least important one. Here is what the numbers actually look like.

| Scenario | n8n Cloud | Self-Hosted | |---|---|---| | Low volume (under 2,500 executions/month) | ~20 to 25 dollars/month | ~20 dollars/month (VPS) | | Medium volume (10,000+ executions/month) | ~50 to 65 dollars/month | ~20 to 40 dollars/month (VPS) | | High volume (100,000+ executions/month) | Pro or Enterprise pricing, climbs fast | ~40 to 80 dollars/month (bigger VPS) | | Setup time | Minutes | 30 to 60 minutes | | Ongoing maintenance | None (n8n handles it) | 2 to 4 hours/month | | Who owns uptime | n8n | You |

The pattern: at low volume the two are roughly even, and Cloud is arguably the better value once you price in your own time. Self-hosting only pulls clearly ahead on cost at high volume, where Cloud's per-execution pricing starts to climb and a flat VPS bill does not.

If you are choosing on price alone at a 40-person firm, it is close to a wash. Which is exactly why price should not be your deciding factor.

This is where most firms reading the book actually make the call.

When you run n8n Cloud, your data passes through n8n's servers. Lead details, client emails, CRM CRMCustomer Relationship Management software. The system of record for contacts, deals, and client communication. Examples: HubSpot, Salesforce, Pipedrive. records, anything your workflows touch, it all travels across infrastructure you do not own. n8n is a reputable company with solid security, and for the majority of businesses this is a complete non-issue.

But if you run a law firm, a financial advisory practice, an RIA, or anything healthcare-adjacent, you likely have obligations that change the math. Attorney-client privilege, financial confidentiality rules, and data residency requirements all push in the same direction: keep client data on infrastructure the firm controls.

Self-hosting answers this cleanly. When n8n runs on your own VPS, client data never leaves a server you own and can point to in an audit. You choose the data center region. You control who has access. You can put it behind your own firewall. When a client or a regulator asks where their data lives, the answer is "on our server," not "on a third-party SaaS platform in another country."

This is not theoretical. It is the single most common reason firms in regulated industries self-host. If you are in one of these verticals, the data residency question usually settles the debate before cost ever enters it. See the compliance notes for law firms, financial advisory, and healthcare-adjacent firms for the specifics in your space.

The maintenance question separates firms that succeed with self-hosting from firms that abandon it.

With n8n Cloud, there is no maintenance. Updates happen automatically. Backups are handled. If the server has a problem, n8n fixes it. You log in and your workflows are running.

With self-hosting, you own the server. That means:

• Updates: n8n ships new versions regularly. You apply them, which is usually a single command or a click depending on your setup.
• Backups: You need a backup of your workflows and credentials so a server failure does not wipe out everything you built. The n8n backup and disaster recovery guide covers this.
• Security: You keep the server patched and locked down. The n8n security hardening guide walks through it.
• Uptime: If the server goes down, your automations stop until you bring it back. For most firms that is a minor inconvenience, not an emergency, but you should know it is on you.

None of this is hard. All of it is recurring. The firms that thrive self-hosting are the ones that name an owner up front, whether that is an internal IT person, an office manager who is comfortable following a runbook, or an outside managed services provider. The firms that struggle are the ones where nobody owns it and the instance quietly falls three versions behind.

Run through these in order. The first one that gives you a clear answer is your answer.

1. Do you handle regulated or highly confidential client data? If yes (law, finance, healthcare-adjacent, anything under strict residency rules), self-host. Data control settles it. Stop here.

2. Do you have anyone who can own a server? This can be an internal IT person, a technically comfortable office manager working from a runbook, or an outside provider. If yes, self-hosting is on the table and usually the right call. If no, lean toward Cloud.

3. How fast do you need to start? If you need to launch your first Play this week and have no technical help, n8n Cloud gets you building in minutes. You can always migrate to self-hosted later by exporting your workflows.

4. What volume do you expect? If you plan to run high volume across many Plays, self-hosting avoids per-execution pricing and stays a flat cost. At low volume, this barely matters.

For most firms reading The AI Workforce Playbook, the answer comes from question 1 or question 2. If you are in a regulated vertical, self-host. If you have no technical resources and your data is not sensitive, start on Cloud.

One reason this decision is lower-stakes than it feels: n8n lets you move. Every workflow exports as a JSON file you can import into any other n8n instance. If you start on Cloud and later decide to self-host, you export your workflows, stand up the new server, reconnect your credentials, and repoint your webhook URLs. Plan a weekend, test each Play on the new instance before you shut the old one down, and you are done.

So make the call that fits where you are today. Start on Cloud if that gets you moving. Self-host if your data demands it. Either way, the workflows you build come with you.

n8n Cloud and self-hosted n8n run identical software. The choice is not about features. It is about who owns the server underneath.

Choose Cloud for speed and zero maintenance when your data is not sensitive and you have no technical help. Choose self-hosted for data control, flat-cost scaling, and full ownership when you handle regulated client information or have someone who can keep a server running.

For the firms this book is written for, data control usually wins, which is why self-hosting is the default recommendation throughout. But there is no wrong answer if you match the choice to your firm honestly. Pick the one you can actually live with, then go build Play 1.