n8n costs for marketers: executions, AI calls and plans

#01n8n plans (checked 4 November 2026)

According to the n8n pricing page, billed annually:

The page lists AI Assistant credits per plan (1,600 for Starter, up to 9,600 for Pro) and a 50% Business discount for startups with under 20 employees. The trial covers 1,000 executions over 14 days. Each plan's base tier includes a monthly execution allowance: 2,500 on Starter, 10,000 on Pro and 40,000 on Business, with unlimited steps per execution. Higher tiers are available.

Plan limits on workflow duration matter for AI workflows: according to the n8n docs, Starter allows up to 5 minutes and Pro up to 40 minutes per execution. Saved executions and log retention also differ (2,500 and 7 days on Starter; 25,000 and 30 days on Pro).

#02Design choices that save money

• Webhooks over polling. Forms, CRMs and many tools can call n8n when something happens.
• Batch work. One scheduled run processing 50 items is one execution; 50 webhook calls are 50.
• Dedupe before paid calls. Don't enrich or summarise the same item twice.
• Use narrow AI nodes and short outputs. Tokens are the main AI cost.
• Cache. Store enrichment and AI results in a data table and reuse them.
• Sub-workflows. Check how your plan counts them before splitting heavily.

#03Self-hosting costs

The Community Edition has no licence fee under the Sustainable Use License, but you pay for a server, a database for larger setups, backups, monitoring and someone's time for upgrades. Lesson 20 covers operations. Execution pruning is on by default with a maximum age of 336 hours (14 days), which keeps the database smaller.

Our view: for most marketing teams, people time is the real cost. A few reliable workflows with alerts beat many fragile ones, whichever plan you choose.

#04Estimating executions

An execution is one run of a workflow, however many nodes or items it has. Rough monthly counts:

Polling triggers run even when nothing is new. A form webhook with 300 submissions a month costs about 300 executions; a 5-minute poll costs about 8,700 whether or not anything changed.

#05Other costs

#06A worked cost estimate (illustrative)

Take a small marketing setup: lead routing by webhook (about 400 runs a month), a weekly report (about 4), a daily competitor monitor (about 30), a weekly AI visibility monitor (about 4) and a content repurposing trigger (about 8 posts). That's roughly 450 executions a month, well within most plans.

AI calls are another matter. If the visibility monitor sends 30 prompts to three engines twice weekly, that's 180 calls a week, plus classification calls for 400 leads a month. Price these with your provider's current per-token rates, using your average prompt and answer lengths.

#07Reducing AI costs

Review actual usage in your provider's dashboard monthly and compare with your estimate.

#08Cost math from the pricing page

Using the n8n pricing page checked 4 November 2026 (billed annually, EUR):

Business costs more per execution because you're paying for self-hosting rights and features such as SSO and environments, not just volume. The pricing page says annual billing saves 17% compared with monthly, and that Starter and Pro are capped with no overage fees, while Business overage is listed at €4,000 per extra 300,000 executions (shown on the page as changed on 26 October 2026).

#09Worked estimate: Ledgerline's plan choice

Ledgerline is a fictional accounts-payable software company. Details are illustrative.

That fits inside Starter's 2,500 with room for growth. The deciding factors were features: Pro's longer execution duration (40 minutes versus 5) mattered for the AI visibility loop, and more concurrent executions helped during webinar sign-up spikes. They chose Pro.

#10The polling trap, in numbers

The pricing page's own example: a workflow that runs every 5 minutes uses about 8,600 to 8,900 executions a month. On its own, that exceeds Starter's 2,500 and uses most of Pro's 10,000. One badly chosen interval can decide your plan.

#11Monitoring usage

The pricing page says you can review production execution counts in the Insights dashboard. Check monthly, and compare with your estimate. A workflow using far more executions than expected usually has a polling trigger or a loop that calls sub-workflows per item.

#12Worked example: Ledgerline's cutting executions

Ledgerline is a fictional accounts-payable software company. Details are illustrative.

Ledgerline's first setup polled HubSpot every 5 minutes for new leads, using about 8,700 executions a month for roughly 400 leads.

Switching to a HubSpot-triggered webhook dropped that workflow to about 400 executions. The AI visibility monitor ran weekly with batching, so it used only a handful of executions; most of its cost was model calls. The team stayed on their plan with room to grow.

Walkthrough

1. List your planned workflows with trigger type and frequency.
2. Estimate executions per month for each using the table above.
3. Estimate AI calls per run and the average input and output length; price using your provider's current pricing page.
4. Add third-party API costs per call.
5. Compare the total with plan limits on the n8n pricing page, including concurrent executions and max duration.
6. Find the biggest cost and see if a webhook, a lower frequency or dedupe reduces it.

#13Try it yourself

Estimate monthly executions and AI calls for three workflows you plan to build. Find one that polls more often than it needs to and redesign it with a webhook or lower frequency.

#14Quick quiz

1. Is an execution counted per node?

Answer: No. One run of a workflow is one execution.

2. About how many executions does a 5-minute poll use per month?

Answer: About 8,600 to 8,900.

3. Are AI model costs included in executions?

Answer: No. They're billed by the provider or through Gateway credits.

#15What should you check before the next lesson?

• Executions are estimated per workflow.
• AI and API costs are estimated per call.
• Polling workflows have been reviewed.

Next: Self-hosting and running n8n in production. Or return to the n8n for Marketers track.

#16Common questions

Is self-hosting free? The Community Edition has no licence fee, but you pay for hosting and maintenance time.

What happens if I exceed executions? The pricing page says Starter and Pro are capped with no overage fees; Business overage is listed per extra 300,000 executions.

Which plan should a small team start on? Start with the trial and estimate executions; the right plan depends on volume and features.