What does a realistic founder workload cost on each?

At 1,000 runs a month of a five-action workflow, Zapier is $1,068 a year and n8n Cloud is €240. The gap is real, and it is smaller than what the wrong automation costs you in hours.

Take a lead-intake flow, as an illustration rather than our data. A form submission triggers it. Then five actions: enrich the lead, write a CRM row, add the contact to your email tool, ping Slack, log it to a sheet. The trigger is free on Zapier, so that is five tasks per run.

Here it is at three volumes:

Same 2,000 runs a month. Zapier's bill follows the number of actions, n8n's does not. Source: zapier.com/pricing and n8n.io/pricing, checked 22 September 2026.

Double the volume and Zapier's bill rises by $480 a year while n8n's does not move. Double the steps as well and Zapier reaches $2,268 a year for the same 2,000 runs. That is the structural truth of the two models: Zapier taxes complexity, n8n taxes frequency.

Now the napkin math that matters more. At 1,000 runs a month, the difference is roughly $800 a year before currency conversion. If your hour is worth $150, that is about five hours a year. One afternoon lost to debugging a self-hosted server, or one silently broken workflow nobody noticed, wipes out the saving. Price the tool by the hours it costs you, not only the invoice.

Can you self-host n8n for free, and what does the license allow?

Yes, for your own internal business use. No, if your customers get to build or edit workflows on it.

n8n is not open source in the strict sense. Its code ships under the Sustainable Use License, which lets you use and modify it "only for your own internal business purposes or for non-commercial or personal use." Files marked .ee sit under a separate enterprise license and need a paid key.

n8n's license FAQ draws the line clearly. Allowed: running your business on it, building automations for clients on your instance as long as they cannot create or edit them, charging for setup and maintenance, and using it behind the scenes in your product. Not allowed: hosting n8n as a service where clients build workflows, letting end users configure workflow logic through your UI, API or an AI agent, and white-labeling it.

For almost every founder reading this, the license is a non-issue. It becomes an issue only if "automation" is the feature you sell. Then talk to n8n's sales team before you build, not after.

Free also does not mean costless. Zapier's comparison makes the fair point that someone has to handle hosting, security patches, backups and monitoring. If that someone is you, the free plan is paid for in founder hours.

Which tasks should a founder automate at all?

The frequent, rule-based ones where a wrong output is cheap. Everything else stays human, whatever tool you pick.

The tool debate skips the harder question. An automation built on the wrong task fails on either canvas. Before you open either canvas, run the task through three filters:

1. Frequency. The classic xkcd chart on automation puts it in numbers over a five-year horizon: shaving five minutes off a daily task is worth up to six days of effort. Shaving five minutes off a monthly task is worth about five hours. Build for the daily ones.

2. Stable rules. If you can write the steps down and they are the same every time, it is a pipeline. If each case needs a judgment call, it is not ready.

3. Cost of a wrong answer. Automate work whose failure mode is a wasted send. Keep a human on work whose failure mode is a burned relationship.

That third filter is the one we live by. In our own outbound pipeline, 332 of 427 companies contacted went to an automated sequence and 95 stayed as drafts that do not send until a human reads them. The automated path handles reach. The manual path handles the leads where a wrong sentence is expensive. Systems propose, the founder approves.

There is a fourth filter that no pricing page will tell you about: whatever you build will break quietly. Our own scheduled pipeline shipped four bugs that produced plausible output and raised no error, and we caught every one by reading the output rather than by being alerted. That is why automations fail silently, and it is the reason the reporting we run on a weekday schedule stays trustworthy: someone reads it. The same habit keeps our founder reporting stack alive.

So which one should you pick?

Pick by workflow shape and by who maintains it, then spend the saved hours on the tasks that should never be automated.

• Pick Zapier if you have a handful of short workflows, low volume, a core tool only Zapier supports, and nobody technical to own the system.

• Pick n8n Cloud if your workflows run hundreds of times a month, chain more than three or four actions, or call an AI model mid-flow. The per-execution model rewards exactly that.

• Self-host n8n if you have someone who will patch it, a reason to keep data on your own server, and no plan to let customers build workflows.

• Use neither when the job needs state between runs and a real change history. A script on a schedule is cheaper and easier to test.

The cheapest automation is the one you decided not to build. The second cheapest is the one where a founder reads the output every week.