The journal

Guides14 min readBy

n8n Automation Cost: What a Workflow Actually Costs

There is no single price for an n8n workflow. A three-step alert costs a fraction of an AI agent that reads email, checks stock and messages a customer on WhatsApp. Here is what actually moves the number.

n8n Automation Cost: What a Workflow Actually Costs

Key takeaways

  • Meant prices n8n automation builds from $2,000 to $20,000, fixed scope and fixed price. n8n automation cost depends on the number of nodes, the external systems a workflow touches, the error handling it needs and where it is hosted. The platform licence is usually the smallest part of the bill.
  • A single-trigger workflow (one app to another, no branching) sits at the low end of a build. Multi-step workflows with conditional logic and several integrations take longer to design and test, so they cost more.
  • Self-hosting n8n removes the per-execution charge of the cloud plans. It also adds server maintenance, uptime monitoring and security patching, and someone has to own that work.
  • Meant prices AI and automation work from $2,000 to $20,000 as a fixed scope agreed in writing before work starts. It is never billed hourly and never charged as a percentage of ad spend.
  • Workflows that call an AI model for drafting, classification or WhatsApp replies cost more to build and run than deterministic ones. Model calls, retries and prompt testing all add cost.

What Actually Drives n8n Automation Cost

Meant prices n8n automation builds from $2,000 to $20,000, fixed scope and fixed price. There is no flat price for a given workflow within that range. n8n automation cost has two parts: the build fee, which depends on how much logic you need, and the running cost, which depends on hosting and on the paid services the workflow calls. At Meant we quote AI and automation work across that range. Where a job lands comes down to five things you can check against your own process today.

The Five Cost Drivers: Nodes, Triggers, Integrations, Logic and Hosting

  1. Nodes. Each step in n8n is a node: fetch a record, reshape a field, send a message. More nodes means more to build, test and document.
  2. Triggers. A webhook from a form is simple. A schedule that polls an inbox, or a workflow started by several different events, needs more care so it does not fire twice or miss a run.
  3. Integrations. Every external system brings its own login method, rate limits and data format. Connecting three systems is not three times the work of one. It is closer to getting three strangers to agree on a date format.
  4. Logic. Branches, loops, deduplication and error handling are where the build time goes. The happy path is quick to build. Deciding what happens when the CRM is down at 2am takes longer.
  5. Hosting. You can run n8n Cloud or your own server. This choice moves cost from the build into the monthly bill, or the other way round.

Why 'How Much Does n8n Cost' Is the Wrong Question

n8n pricing, meaning the licence, is often the cheapest line in the whole project. The self-hosted edition costs nothing, and the cloud plans start in the tens of euros a month. The money goes into design: deciding what should happen to each record in each case, then proving that it does. A better question has three parts. What does this specific workflow need to do? How often does it run? What breaks for the business if it fails?

n8n Cloud Pricing vs Self-Hosting: The Decision That Changes the Number

Hosting is the one decision that changes the shape of the bill. Cloud gives you a predictable subscription with limits. Self-hosting gives you no usage limits but adds an operational job.

n8n Cloud: Paying by Execution

n8n Cloud is billed by execution. The Starter plan is €24 a month for 2,500 executions, Pro is €60 a month for 10,000, and Business is €800 a month for 40,000 executions with single sign-on. Enterprise is priced on request. Annual billing saves 17% on all cloud plans (source).

An execution is one complete run of a workflow, however many steps it has. A twelve-node workflow that runs once uses one execution.

The limit that catches people out is run time. Cloud Starter stops a single run after five minutes and runs five workflows at once. Pro raises the time limit to 40 minutes (source). A batch job or an AI agent waiting on a slow model can hit the Starter ceiling.

Self-Hosted n8n: No Per-Execution Fee, But Someone Owns the Server

The Community Edition is free to self-host, with unlimited workflows, unlimited executions and no feature gating. At 20,000 executions a month, self-hosting costs roughly $10, against $60 to $100 or more on managed platforms (source).

The server is cheap, but looking after it is not free. Someone has to apply n8n updates, back up the database, renew certificates, patch the operating system and notice when the box goes down. That labour cost varies by team and region, so there's no single figure to quote here, only the fact that the job needs an owner. If it has none, the first sign of trouble is usually a sales lead that never arrived.

If your volume grows well past these examples, say into the hundreds of thousands of executions a month, the hosting decision becomes the dominant line in your total cost, well ahead of any per-message or per-token fee discussed below.

Which Route Suits a Small Team vs a Growing One

A small team with a few workflows and no in-house technical person is usually better on Cloud. The subscription costs less than the attention a server needs. A growing business with high volumes, long-running jobs or data that must stay on its own infrastructure tends to move to self-hosting, provided someone is named as the owner.

n8n CloudSelf-hosted n8n
LicenceIncluded in planFree (Community Edition)
Monthly costFrom €24 (Starter) to €800 (Business)Server cost only, roughly $10 at 20,000 executions
Execution limits2,500 to 40,000 depending on planUnlimited
Run time per execution5 minutes on Starter, 40 on ProSet by your server
MaintenanceHandled by n8nYours: updates, backups, security, uptime
Best forSmall teams, modest volumeHigh volume, long runs, data control

What a Simple Workflow Costs

Example: A Form Submission That Creates a CRM Record

A visitor fills in the contact form on your site. The form sends a webhook to n8n. n8n maps the fields, creates a contact in your CRM and posts a notification to the sales channel. That is four or five nodes, one trigger and no decisions. The volume fits comfortably inside Cloud Starter.

This is the kind of job that sits at the start of our range, from $2,000, as a fixed price. Drawing the nodes takes a fraction of that time. Most of the work is testing with real submissions, handling odd inputs such as a phone number typed with spaces, and making sure a failure is visible to someone.

What Keeps This Tier Cheap

  • One system sends and one system receives.
  • The data arrives in a predictable shape.
  • There is no branching: every record follows the same path.
  • A failure is low-stakes and easy to spot.
  • You already have admin access to both tools and their APIs.

If this describes everything you need and you are comfortable in n8n, you can probably build it yourself. We would rather say so than sell you a build you do not need.

What a Multi-Step Integration Costs

Example: Lead Capture, CRM Sync and a WhatsApp Follow-Up

This is where most real business workflows sit. Here is one example:

  1. A lead arrives from a Meta lead ad or a site form.
  2. n8n checks the CRM for an existing contact, so it does not create duplicates.
  3. It creates or updates the record and assigns an owner by product or region.
  4. It sends an approved WhatsApp template message through the Business API.
  5. If there is no reply within a set window, it sends an email instead.
  6. It logs the outcome back to the CRM.

CRM sync means the same record stays correct in both systems without anyone copying data across by hand. If the CRM side is the larger part of your project, we have written separately about what a CRM integration costs on its own.

The running cost now includes Meta's own fees, which n8n does not absorb. For US and Canada numbers, since 1 October 2026, Meta charges $0.025 per marketing message and $0.0034 per utility, authentication or service message. Each phone number gets 1,000 free service messages a month. Rates vary by country: the same ordering workflow that costs about $42 a month in Meta fees in the US costs about $136 in the UAE. At higher sending volumes these per-message fees scale linearly with messages sent, so a business sending tens of thousands of WhatsApp messages a month should budget Meta's fees as their own line item, separate from the n8n build and hosting cost above.

Why Branching Logic and Error Handling Add Build Time

Every branch is another path to test. Six branches is six sets of test data, six ways to fail and six things to document.

Error handling is the part that clients do not see and later rely on. It includes:

  • retries with a pause between attempts when an API is busy
  • a separate error workflow that alerts a named person
  • a list of failed records that can be replayed once the fault is fixed
  • guards that stop the same lead from getting two WhatsApp messages if the form fires twice

None of this shows up in a demo. All of it shows up the first time a third-party API has a bad afternoon. Builds like this sit in the middle of our range.

What an AI Agent Workflow Costs

Why AI Nodes Cost More to Build and Run

A deterministic workflow gives the same output for the same input, so you test it once and move on. A model does not behave that way. Building around one means:

  • testing prompts against a set of real messages
  • defining what the model must never say
  • checking that its output is in the format the next node expects, and retrying when it is not
  • deciding the point at which a human takes over
  • logging enough detail to see why it answered the way it did

Running costs are simpler than many people expect. n8n adds no AI surcharge, so the total is hosting plus the model provider's API fees. GPT-4o costs around $5 per million input tokens and Claude Sonnet around $3 (source). Run time matters too: an agent that makes several model calls in one run can exceed the five-minute limit on Cloud Starter. As with WhatsApp messaging, token costs scale with the volume of conversations the agent handles, so a high-traffic agent should be budgeted on expected monthly conversation count, not just the per-token rate.

Example: An Agent That Reads, Drafts and Replies on WhatsApp

  1. A customer message arrives on WhatsApp.
  2. A model classifies the intent: order status, booking, complaint or other.
  3. n8n pulls context from Shopify or the booking system.
  4. The model drafts a reply using that context.
  5. A confidence check decides whether the reply is sent or passed to a person, with the draft attached.
  6. Everything is logged.

Builds like this sit at the upper end of the range because of the testing involved.

There is an alternative. Meta's own Business Agent Platform launched on 1 July 2026 and answers WhatsApp users automatically. It is billed per token at $2.00 per million tokens, which works out to roughly 4 to 5 cents for a typical message. It takes less setup. The trade-off is less control over what it can look up, how it uses your own systems and when it hands a conversation to a person.

n8n vs Make: Does the Platform Itself Change the Price

Where the Two Platforms Price Similarly

The build fee is driven by the same five factors on either platform. Both are visual builders, and both connect to the common CRMs, inboxes and messaging tools. For a simple workflow, the design and testing time is about the same whichever platform you choose.

Where They Diverge

The difference is in the running cost, and the two platforms bill on different units entirely. n8n counts one execution per complete run, however many steps are inside it. Make counts credits, and its tiers set the cost directly: the Free plan gives 1,000 operations a month, Core runs $9 a month (billed annually) for 10,000 credits, Pro is $16 a month (annually) for 10,000 credits plus priority execution, and Teams is $29 a month (annually) for 10,000 credits plus team roles, with Enterprise on a custom quote (source). Because Make spends credits per step rather than per run, a ten-step workflow run many times a day will burn through a 10,000-credit plan faster than the equivalent n8n workflow uses its execution allowance. Only n8n can also be self-hosted. Make is easier for a non-technical team to maintain. For the full picture, read a full comparison of Make and n8n.

n8nMake
Billing unitOne execution per full runCredits, spent per step in each run
Entry priceFree self-hosted; Cloud from €24/monthFree plan, then from $9/month (Core, annual)
Long workflows at volumeCost stays flat per runCredit use grows with step count
Self-hostingYes, free Community EditionNo
Custom codeJavaScript and Python code nodesBuilt mainly around prebuilt modules
Easiest forTeams with some technical ownershipNon-technical teams maintaining it

How to Budget for Your Own n8n Workflow

Questions to Answer Before You Ask for a Quote

  1. What starts the workflow: a form, an email, a schedule, a message?
  2. Which systems does it touch, and do you have admin access and API keys for each?
  3. How many times a month will it run, now and in a year?
  4. Where does it make decisions, and what are the rules?
  5. What should happen when it fails, and who needs to know?
  6. Does any step need an AI model?
  7. Who will own hosting and maintenance?

With those answers you can place your workflow in the simple, multi-step or AI tier above. You can also calculate the payback time on an automation build against the hours it saves. On one build, that saving came to 20 hours a week.

Getting a Fixed Price Instead of an Estimate

An estimate tells you where the bill might start. A fixed price tells you where it ends. We work out the scope on a free 30-minute call, then put the scope and price in writing before any work starts. That is how we quote automation work as a fixed price. Ongoing support runs as a month-to-month retainer that you can stop. If the honest answer is that you can build it yourself on Cloud Starter, we will tell you on the call.

Frequently asked questions

Is n8n free to use for business automation?

The self-hosted Community Edition is free, with unlimited workflows and executions. You still pay for the server and for someone's time to maintain it. n8n Cloud starts at €24 a month for 2,500 executions.

How long does it take to build an n8n workflow?

Timelines vary by scope rather than following a fixed figure. A single-trigger workflow with no branching is a short job. A multi-step build with several integrations, branching logic and error handling takes longer, mostly because of testing edge cases and waiting for API access from third-party systems. We agree the actual timeline in writing alongside the fixed price, once we know the scope.

Can n8n connect to WhatsApp Business API directly?

Yes. n8n has a WhatsApp Business Cloud node and can also call the API directly over HTTP. You need a Meta Business account, a registered number and approved templates for messages that open a conversation. Meta bills its per-message fees separately.

Do I need a developer to build n8n workflows, or can I do it myself?

A simple workflow that connects one app to another is well within reach of a confident non-developer. Once you add several systems, branching, error handling or an AI model, a developer earns their fee. Those workflows tend to fail quietly, and a quiet failure costs more than the build.

What happens if an n8n workflow breaks after it's live?

A well-built workflow has an error workflow that alerts a named person and keeps the failed records so they can be replayed. Common causes are an expired credential, a changed API or a renamed field in the CRM. We fix these under a month-to-month retainer, or we hand over documentation so your team can.

If you want a fixed price on your own workflow instead of a guess, book the free 30-minute call with Meant. We will tell you what the build costs, what it will cost to run each month, and whether you need us at all.

Get the next one.

One email when a new case study or guide goes up. No list, no newsletter filler.

No list, no filler. Reply to any entry and we will take you off.

Next article

App Development Timeline: How Long a Build Really TakesGuides
WhatsApp us