n8n pricing is easiest to understand when you stop counting nodes and start counting workflow runs. Under n8n’s current Cloud pricing, the principal usage unit is a workflow execution: one complete run of a workflow counts as an execution even if that run contains many steps. That can make long workflows behave very differently from platforms that meter each successful action.
As of August 31, 2026, n8n lists Starter at €20 per month billed annually for 2,500 executions and Pro at €50 per month billed annually for 10,000 executions. n8n also lists Business as a paid self-hosted tier and Enterprise with custom usage and hosted or self-hosted options. Pricing changes, so use the official pricing page for a final purchase decision.
What exactly is an n8n execution?
An execution is a run of a workflow. A webhook that fires once and moves through twelve nodes is one execution under the current Cloud model; a scheduled workflow that runs every five minutes can accumulate many executions even if it contains only three nodes. Frequency therefore matters more than visual length when estimating base Cloud usage.
Count production runs, not workflow files. One workflow can run thousands of times per month while ten rarely used admin workflows contribute little usage.
Estimate scheduled workflows with simple arithmetic
Scheduled jobs are the easiest to forecast. A workflow that runs hourly executes about 24 times per day. A workflow that runs every 15 minutes executes about 96 times per day. Multiply by the number of active days and allow headroom for retries, manual tests, or extra schedules.
| Schedule | Approx. runs/day | Approx. 30-day runs |
|---|---|---|
| Once daily | 1 | 30 |
| Hourly | 24 | 720 |
| Every 30 minutes | 48 | 1,440 |
| Every 15 minutes | 96 | 2,880 |
| Every 5 minutes | 288 | 8,640 |
These are planning estimates, not billing guarantees. Actual executions depend on how the workflow is configured and run.
Estimate event-driven workflows from business volume
For webhooks and app events, start with the business event count: leads per month, orders per day, tickets created, forms submitted, files uploaded, or records changed. Then ask whether each event triggers one workflow or several.
A company receiving 1,200 leads per month might use one execution per lead for routing, while a second follow-up workflow runs only for qualified leads. Model those separately instead of assuming every event follows the same route.
Workflow length matters operationally even when it does not drive the base execution count
A 25-node workflow may still create cost in external APIs, model tokens, database operations, and maintenance time. The n8n billing unit does not erase downstream vendor charges. An OpenAI step, enrichment provider, SMS service, or paid data API can dominate total cost even when n8n itself counts one execution.
Estimate the entire workflow economics: n8n plan, third-party usage, human review, and infrastructure if self-hosted.
Starter: who 2,500 executions can fit
Starter is easiest to fit when workflows are relatively low-frequency: daily reports, a few hundred monthly leads, periodic data synchronization, or operational automations that do not trigger continuously. One 15-minute scheduled workflow alone can exceed 2,500 executions in a typical month, which illustrates why schedule frequency deserves attention.
The relevant question is whether expected production run volume plus growth and testing sits comfortably below the plan limit—not whether the team has only a handful of workflows.
Pro: when 10,000 executions and collaboration become more relevant
Pro’s larger execution allowance can fit teams with several event-driven workflows or more frequent schedules. Current n8n pricing also positions paid Cloud plans with collaboration/workflow capabilities that should be checked directly for the features your team needs.
Do not upgrade purely because the workflow count grows. Upgrade when execution volume, required features, or collaboration needs justify it.
Business and Enterprise change the deployment conversation
n8n currently positions Business as a paid self-hosted plan and Enterprise for custom usage with hosted or self-hosted options. Buyers at this level should evaluate governance, support, environments, security requirements, and operational responsibilities—not just execution volume.
A self-hosted paid plan still means your organization owns infrastructure duties. Compare Cloud vs self-hosted before assuming a higher plan is only a feature decision.
What does “free” mean in the pricing model?
n8n’s Community Edition provides a long-term self-hosted path under the current license, while n8n Cloud uses a trial rather than a permanent free Cloud tier under the current published model. Self-hosting avoids a normal Cloud subscription for permitted Community Edition use but does not make infrastructure, backups, upgrades, or staff time free.
See Is n8n free? for the practical difference between Community Edition and Cloud trial access.
Example budget: lead operations
Assume a business receives 1,500 leads per month. Each lead triggers one intake workflow. Roughly 40% of leads qualify and trigger a second enrichment-and-routing workflow. A daily summary runs once per day.
- Lead intake: about 1,500 executions.
- Qualified-lead workflow: about 600 executions.
- Daily summary: about 30 executions.
- Base estimate: about 2,130 executions before tests, retries, growth, or other workflows.
This example demonstrates the method, not an n8n invoice. Your architecture can combine or split workflows differently, which changes execution counts.
Example budget: frequent polling can dominate usage
A single workflow scheduled every 10 minutes runs roughly 4,320 times in a 30-day month. If the poll often finds nothing, those executions may still be a poor use of allowance compared with a webhook or less frequent incremental sync.
Pricing analysis can therefore reveal architecture improvements. The goal is not to game billing; it is to avoid unnecessary work.
Cloud cost versus self-hosted cost
Self-hosting replaces part of the subscription question with an operations question. Include compute, database, backups, monitoring, TLS, upgrades, security, incident response, and staff time. A small server bill does not represent total cost if an engineer spends hours each month maintaining the platform.
Conversely, an organization that already operates production infrastructure and has a real private-network requirement may value self-hosting for reasons that are not captured by the Cloud plan table.
Pricing FAQ
Does each node count as an execution?
Under the current n8n Cloud model, billing centers on workflow executions rather than counting every node as a separate execution. Verify the current pricing definition before purchase.
Do failed runs count?
Billing and execution-count details can change and may depend on product rules. Check n8n’s current pricing/usage documentation for how your specific runs are counted.
How much headroom should I plan?
Use enough headroom for growth, tests, manual reruns, retries, and new workflows. The exact percentage depends on how predictable your business volume is.
Is self-hosted n8n free?
Community Edition can avoid a standard software subscription for permitted uses, but infrastructure and operations still cost money and license boundaries still apply.
What is the best way to estimate a plan?
List every production workflow, estimate monthly executions from schedules or business events, then add collaboration/features and external service costs.
Build a monthly execution model before comparing plans
The cleanest pricing estimate starts with workflow frequency rather than the number of workflows in the editor. Create a simple inventory with one row per production workflow and four fields: trigger type, expected runs per day or month, growth factor, and whether downstream events start additional workflows. A daily finance digest may contribute only about thirty runs in a month, while a polling workflow scheduled every five minutes can contribute thousands. The second workflow is the one that drives plan pressure even if it contains fewer nodes.
Event-driven automations need a different estimate. Start from business volume: orders, leads, tickets, form submissions, uploaded files, or records changed. If 3,000 orders arrive each month and every order triggers one workflow, that is a base of roughly 3,000 executions before follow-up workflows, manual replays, testing, and exceptional paths. If only 15% of those orders trigger a separate fraud-review workflow, model that additional 450 runs separately instead of assuming every order takes the same route.
Budget for retries, replays, and operational headroom
A production estimate that uses exactly 100% of a plan allowance is fragile. Workflows are tested, manually rerun, changed, and occasionally retried after upstream failures. Traffic also tends to be uneven. A campaign launch, month-end process, product release, or seasonal event can create a temporary spike that a flat monthly average hides. Build headroom into the estimate so ordinary operations do not require constant plan management.
Failure handling can also change execution behavior depending on how you design the workflow. A failed request might be retried inside one workflow, routed to a separate error workflow, or replayed manually after the underlying issue is fixed. Because architecture affects usage, the pricing worksheet should be reviewed whenever you materially change retry strategy, polling frequency, or workflow boundaries.
Separate n8n cost from the rest of the automation bill
An n8n subscription is only one line in total automation cost. A workflow may also consume paid API calls, AI tokens, enrichment credits, email or SMS usage, database capacity, proxy services, storage, observability tools, or staff time. A long workflow can still be inexpensive in n8n execution terms while being costly because one downstream service charges per record or request.
For self-hosted deployments, replace some subscription cost with infrastructure and operations. That can include compute, a database, backups, monitoring, TLS termination, secrets management, upgrades, and engineer time. The correct comparison is therefore not “Cloud price versus free software.” It is “managed platform cost versus the full cost of operating the required deployment yourself.”
Three usage profiles that produce very different bills
Low-frequency operations
A small team may run a nightly report, a few hundred monthly lead automations, and occasional administrative workflows. Even with several workflow files, execution volume can remain modest because triggers are infrequent. Here the deciding factor may be convenience and collaboration features rather than raw allowance.
High-frequency polling
A single scheduled process can dominate usage. Polling every ten minutes produces roughly 4,320 scheduled runs in a 30-day month before the workflow does anything useful. If the source supports webhooks, replacing polling with event delivery can materially change the plan estimate and reduce empty work.
High-volume event processing
Ecommerce, support, product telemetry, or lead systems may produce thousands of real events. The economics then depend on whether one event maps to one workflow execution or fans out across several independent workflows. Draw that fan-out explicitly before choosing a plan.
A buyer worksheet for n8n pricing
- List every production trigger and its expected monthly frequency.
- Estimate secondary workflows started by those runs.
- Add a realistic allowance for testing, retries, replays, and growth.
- Identify third-party costs that scale per request, message, record, or token.
- Decide whether Cloud convenience or self-hosted control better matches the team.
- Recheck the official n8n pricing page immediately before purchase, because plans and included features can change.
This method prevents two common pricing mistakes: choosing a plan from workflow count alone, and assuming self-hosting makes the overall system costless. The execution model is useful precisely because it ties the plan decision to how the business process actually runs.
Final recommendation
Price n8n from workload shape. Count scheduled runs and business events, model separate workflows independently, add realistic headroom, and include external APIs or self-hosted operations. The execution model can be attractive for long workflows, but frequent schedules can consume allowances quickly. Verify the current plan table before committing.
Pricing and plan features can change. The first-party sources below are the current references used for this analysis.
Sources & verification
Product facts checked August 31, 2026. Always verify current vendor terms before purchase or deployment.