How Many n8n Executions Do You Need Per Month?

n8n Cloud pricing becomes easier to predict once every trigger is converted into monthly workflow runs. Count schedules, webhook events, and AI messages separately, add them together, then leave headroom for growth and retries.

Commercial note: links through loke.dev/go are measured as outbound clicks. The n8n link is currently direct and earns no commission. If that changes, the relationship label and disclosure will change.

If you already know your daily volumes, use the free n8n execution and hosting cost calculator. The method behind it is explained below.

What n8n calls an execution

The current n8n pricing page defines an execution as one complete run of a workflow. A workflow with two steps and one with fifty steps each use one execution when they run once. The amount of data processed inside that run does not change the execution count.

That is different from automation products that count each operation or task. It also means node count is not the right input for your estimate. Trigger frequency is.

Use this formula for a first estimate:

monthly executions = scheduled runs + webhook events + chat messages + other production triggers

The result is a capacity estimate, not a billing guarantee. Check the usage shown in your n8n account after real workflows have been running.

Scheduled workflow reference

For an average month of 30.4 days, one scheduled workflow produces roughly:

• once per day: 30 executions
• once per hour: 730 executions
• every 15 minutes: 2,918 executions
• every 5 minutes: 8,760 executions
• every minute: 43,800 executions

Multiply each value by the number of independent workflows on that schedule. Four hourly workflows are about 2,918 executions per month, not 730.

Use the actual schedule window when a job runs only during business hours or weekdays. A five-minute job that runs for eight hours on Swedish working days is much smaller than a five-minute job running all month.

Webhook and event-driven workflows

For a webhook, the outside event volume is the execution volume. A payment workflow with 20 events per day is about 608 monthly executions. A lead form receiving 100 submissions per day is about 3,040.

Use measured application traffic when it exists. If the product has not launched, make a low, expected, and high scenario instead of pretending one forecast is certain.

• Low: current test or pilot traffic
• Expected: realistic launch traffic
• High: a busy week repeated for a month

Choose a plan from the expected case only when the high case can be handled without breaking a business process. For payments, alerts, and customer requests, headroom matters.

AI chat and agent workflows

n8n’s pricing guidance recommends estimating chatbot usage from conversations and average messages per conversation. If every message starts the workflow, use:

monthly chat executions = conversations per week × messages per conversation × 4.345

For 50 conversations per week with eight messages each, the estimate is about 1,738 executions per month. Tool calls and model tokens have separate provider costs, even when the n8n workflow still counts as a single run.

Check the actual trigger design. A conversation handled in one long workflow behaves differently from a chat trigger that starts a new run for every message.

Three useful examples

Small internal automation

• daily reporting workflow: 30
• CRM webhook at 20 events per day: 608
• weekly cleanup workflow: 4

Estimated total: 642 executions per month.

Several frequent data syncs

• three hourly sync workflows: 2,189
• daily reconciliation workflow: 30
• about 25 webhook events per day: 760

Estimated total: 2,979 executions per month. This passes the current 2,500-execution Starter reference even before headroom.

AI support assistant

• 25 conversations per day
• eight messages per conversation
• two daily maintenance workflows

Estimated total: 6,141 executions per month. Model, vector database, and other API costs are separate.

Add headroom before choosing a plan

I would avoid planning to consume 100% of an allowance. Product launches, retry logic, duplicated webhooks, and new workflows create normal variation. A 20% to 30% planning margin is a reasonable engineering starting point, not an n8n rule.

With 25% headroom:

planned capacity = expected monthly executions × 1.25

A workload estimated at 8,500 executions becomes 10,625 for planning. That changes the decision even though the average month might fit a 10,000-execution reference.

Do not optimize the wrong thing

Combining several items into one workflow run can reduce executions, but batching changes failure handling, latency, memory use, and retry behavior. Do it only when the business operation is naturally a batch.

Likewise, do not merge unrelated workflows only to make the counter smaller. A clear workflow that fails safely is usually worth more than a dense workflow designed around a pricing metric.

Measure after launch

Use the executions view to inspect which workflows run, fail, and get retried. Compare the first full week with the forecast, then update the estimate.

• rank workflows by run count
• check unexpected trigger spikes
• separate normal growth from duplicate events
• review retention and saved-execution limits as well as run volume

Keep the estimate in the same place as the workflow inventory. When someone adds a new five-minute schedule, the monthly capacity decision should be updated before the workflow is published.

Choose from the current numbers

As verified on 24 July 2026, n8n listed Starter at €20 per month billed annually for 2,500 executions and Pro at €50 for 10,000. Those prices and limits can change. Check current n8n pricing before purchasing.

Run your own numbers in the calculator. If the managed plan fits with headroom, try n8n Cloud. If it does not, compare the real self-hosted total, including maintenance and recovery work, before deciding.