How to Build Your First n8n Workflow: A 2026 Guide

Why n8n over Zapier or Make

Most automation tools charge per task and lock you into their cloud. Zapier costs $69/month for 2,000 tasks on their Starter plan. Make (formerly Integromat) is cheaper but still metered. Scale up to tens of thousands of operations per month and you are looking at hundreds or even thousands in monthly fees - for moving data between tools you already pay for.

n8n is open-source, self-hostable, and gives you unlimited executions for free. For businesses building serious automation infrastructure, this matters. You pay for hosting (typically $5-20/month on a VPS) or use n8n Cloud if you prefer managed infrastructure, and your costs stay flat regardless of volume.

We use n8n across every Bloodstone venture - from content pipelines that publish 16 articles a day to CRM workflows that qualify leads before they hit our inbox. It is the backbone of our automation service and the tool we recommend to every client looking to reduce manual work.

Beyond cost, n8n gives you something Zapier and Make do not - full control. You can write custom JavaScript or Python inside nodes, connect to any API with HTTP request nodes, handle complex branching logic, and debug workflows visually. When an automation breaks at 2am, you can see exactly which node failed, what data it received, and what went wrong. Try getting that level of visibility from Zapier.

The anatomy of a workflow

Every n8n workflow follows the same pattern: trigger, process, action. Understanding this pattern is the key to building any automation, no matter how complex.

Trigger - something that kicks off the workflow. This could be a webhook (your website sends data to n8n), a scheduled timer (run every morning at 9am), an incoming email, a new row in a Google Sheet, a new message in Slack, or a database change. The trigger determines when your workflow runs.

Process - one or more nodes that transform, enrich, filter, or route the data. This is where the logic lives. You might clean up form data, call an API to enrich a lead, use an IF node to route different types of requests, or call an AI model to classify or summarise content.

Action - the output. Send an email, write to a database, post to Slack, create a Trello card, update a CRM record, or trigger another workflow. Most workflows have multiple actions - notify the team, log the data, and send a confirmation to the customer.

The power is in the chaining. A single workflow can watch for new form submissions, enrich the data via an API, score the lead using Claude, write the result to Supabase, and send a personalised follow-up email - all in under three seconds.

Your first workflow: form-to-CRM (step by step)

Here is the simplest useful workflow you can build today. We will walk through each node so you can follow along in n8n.

Step 1 - Set up the trigger (Webhook node)

Open n8n, create a new workflow, and add a Webhook node. This generates a unique URL that your website contact form will POST data to. Configure it to accept POST requests and set the response mode to "Last Node" so the form gets a confirmation back.

Copy the webhook URL. In your website's contact form, set the form action to this URL. When someone submits the form, the data arrives in n8n as a JSON object with fields like name, email, message, and whatever else your form collects.

Step 2 - Clean the data (Set node)

Add a Set node after the webhook. This lets you restructure the incoming data into a clean format. Map the fields you need - full_name, email, message, source (hardcode this to "website") and add a timestamp. This step ensures your data is consistent regardless of how the form sends it.

Step 3 - Classify the lead (HTTP Request node)

Add an HTTP Request node and point it at the Claude API (or whichever AI model you prefer). Send the lead's message in the prompt and ask Claude to classify the intent - is this a sales enquiry, a support request, a partnership proposal, or spam? Have it return a structured JSON response with the classification and a confidence score.

This step alone saves hours of manual triage. Instead of reading every form submission to decide who handles it, the AI does it instantly.

Step 4 - Route the lead (IF node)

Add an IF node that checks the classification. If the intent is "sales" and the confidence is above 0.7, route to the sales path. Everything else goes to a general inbox path. You can add as many branches as you need - support requests to your helpdesk, partnership enquiries to a specific team member, spam to the bin.

Step 5 - Store the data (Supabase node)

Add a Supabase node (or your database of choice) to insert the enriched lead into your leads table. Include all the original form data plus the AI classification, confidence score, and routing decision. Now you have a clean, searchable database of every lead with automatic categorisation.

Step 6 - Notify the right person (Email/Slack node)

Add a notification node at the end of each branch. Sales leads trigger a Slack message to your sales channel with the lead details and AI classification. Support requests create a ticket in your helpdesk. The email includes a direct link to the lead record in your database so whoever picks it up has full context immediately.

This six-node workflow replaces what would otherwise be a manual process of checking form submissions, copying data into a spreadsheet, reading each message, deciding who should handle it, and forwarding it to the right person. It runs 24/7 and handles leads in seconds, not hours.

Real business examples

Beyond form processing, here are workflows we have built for clients that show what n8n can do at scale.

Invoice processing. A workflow watches a shared email inbox for incoming invoices. When a PDF arrives, it extracts the attachment, sends it to an AI model for data extraction (supplier name, amount, due date, line items), validates the extracted data against the supplier database, and creates a draft entry in the accounting system. The finance team reviews and approves rather than manually entering every invoice. For one client, this cut invoice processing time from 15 minutes per invoice to under 2 minutes.

Lead routing and enrichment. When a new lead enters the CRM (from any source - website, LinkedIn, referral, event), n8n enriches it with company data from an API, scores it based on custom criteria (company size, industry, location), assigns it to the right salesperson based on territory rules, and sends a personalised welcome email within 60 seconds of the form submission. This is the kind of workflow our automation team builds regularly.

Daily report generation. Every morning at 8am, a workflow pulls data from multiple sources - Google Analytics, Stripe, the CRM, and the support ticket system. It compiles the numbers into a formatted summary and posts it to a Slack channel. The team starts every day with a snapshot of yesterday's performance without anyone having to manually pull reports.

Content pipeline. For our media properties, n8n monitors RSS feeds from dozens of sources. When new articles appear, it filters for relevance, sends the content to Claude for rewriting and categorisation, generates metadata and SEO tags, and publishes to the CMS. This runs twice daily and handles the entire editorial pipeline without manual intervention.

n8n vs writing code

A reasonable question: why not just write a script? If you know Python or Node.js, you could build any of these workflows in code. There are three reasons n8n wins for most business automation.

Visibility. When a workflow fails, n8n shows you exactly which node failed, what data it received, and what the error was. With a script, you are reading log files and guessing. Non-technical team members can look at an n8n workflow and understand what it does. They cannot read a Python script.

Speed of iteration. Changing a workflow in n8n takes minutes. Drag a new node in, connect it, configure it, test it. Changing a script means editing code, testing locally, deploying, and hoping nothing breaks. When business requirements change weekly (and they do), this speed difference compounds.

Maintenance burden. Scripts need dependency updates, server maintenance, error monitoring, and someone who understands the codebase. n8n handles the infrastructure. You focus on the logic.

That said, n8n is not the right choice for everything. If you need sub-millisecond performance, complex data transformations at massive scale, or deeply custom business logic, code is still the answer. For 90% of business automation - moving data between systems, triggering actions based on events, enriching and routing information - n8n is faster to build and easier to maintain.

Self-hosted vs n8n Cloud

You have two options for running n8n, and the choice matters.

Self-hosted means running n8n on your own server. The software is free. You pay for hosting ($5-20/month on Digital Ocean, Hetzner, or AWS). You get unlimited workflows, unlimited executions, and full control over your data. The trade-off is that you are responsible for updates, backups, and uptime. If you have a technical team or a partner managing your infrastructure, self-hosting is the clear winner on cost.

n8n Cloud is the managed service. Plans start at $24/month for limited executions and go up from there. You get automatic updates, built-in backups, and n8n handles the infrastructure. The trade-off is cost at scale - heavy usage on Cloud can get expensive quickly.

Our recommendation: start with n8n Cloud to learn the platform and build your first workflows. Once you have workflows running in production and understand your execution volume, evaluate whether self-hosting makes sense. For most of our clients, we set up self-hosted n8n as part of the automation infrastructure because the long-term cost savings are significant.

Common mistakes to avoid

Over-engineering early. Start with one workflow that solves one real problem. Don't try to automate your entire business on day one. Get a single workflow running reliably, prove the value, then expand. We have seen companies spend weeks designing an elaborate automation architecture before building anything. Start simple. Iterate.

Ignoring error handling. Every n8n workflow should have an error trigger that alerts you when something breaks. Add an Error Trigger node to every workflow and connect it to a Slack or email notification. Silent failures are the enemy of automation trust. If your team does not trust the automation, they will keep doing things manually, and the whole exercise was pointless.

Not versioning. Export your workflows as JSON and store them in Git. When something breaks at 2am, you want to be able to roll back to the last working version. n8n does not have built-in version control (yet), so you need to handle this yourself. We export workflow JSON after every significant change and commit it to the project repository.

Skipping testing. n8n lets you test individual nodes and full workflows with sample data. Use this. Test with realistic data, edge cases, and error scenarios before setting a workflow to active. A workflow that works perfectly with your test data but breaks on the first real submission is worse than no automation at all.

Building without monitoring. Once a workflow is live, you need to know it is running. Set up a simple health check - a scheduled workflow that runs every hour and sends an alert if your critical workflows have not executed within their expected timeframe. This catches issues before your team or customers notice.

Next steps after your first workflow

Once your first workflow is running reliably, the natural next steps are:

1. Identify your next highest-value manual process. Look for tasks that are repetitive, time-sensitive, or error-prone. These are your best automation candidates.
2. Connect your core tools. Most businesses run on 5-10 software tools. Map out how data flows between them and identify where manual copying, exporting, or re-entering happens. Each of those is a workflow opportunity.
3. Add AI to existing workflows. Once you are comfortable with basic trigger-process-action workflows, start adding AI nodes. Classification, summarisation, data extraction, and content generation can be added to almost any existing workflow to make it smarter.
4. Build a workflow library. Document your workflows, what they do, when they run, and who owns them. As your automation footprint grows, this documentation becomes essential.

If you are building your first workflow and want guidance on architecture, error handling, or AI integration, get in touch. We build n8n workflows for clients every week as part of our automation service and can get you up and running fast.

Want to see how n8n compares to Zapier? Read our n8n vs Zapier comparison for a full breakdown of pricing, features, and when to use each.