3 n8n Nodes Every Marketer Should Know (And What to Build With Them)

Most marketing automation tools are designed to make simple things easy. n8n is designed to make complex things possible — which is why it's worth the steeper learning curve if you're building an AI-native marketing function rather than just connecting a CRM to a Mailchimp account.

I've been using n8n to build marketing automation for my own pipeline and for client work, and the same three nodes show up in almost everything useful I build. Master these three and you can automate content operations, outbound personalisation, lead enrichment, and multi-step nurture sequences without needing a developer or a stack of expensive point solutions.

This is not a beginner's guide to n8n. It assumes you've set it up (self-hosted or cloud), know how to create a workflow, and have connected at least one integration. If you haven't — start there first.

Why n8n, Not Zapier or Make?

Zapier and Make are excellent for straightforward integrations. If you need to push a new HubSpot contact into a Slack channel or add a Typeform submission to a Google Sheet, either will do the job cleanly. Use them for simple, linear workflows.

n8n's advantage is different. First, it runs code natively — the Code node lets you write JavaScript or Python inside a workflow, so you can manipulate data in ways that Zapier's built-in transforms can't handle. Second, it calls any HTTP endpoint, which means it integrates with any API without needing a pre-built connector. Third, it's self-hostable, which matters if you're handling prospect data under GDPR and don't want it flowing through a US SaaS platform.

For AI-assisted marketing workflows in particular — where you're calling LLM APIs, enriching data on the fly, and chaining logic across multiple services — n8n is materially more capable than the alternatives at similar price points.

Node 1 — The HTTP Request Node

The HTTP Request node is n8n's bridge to the world. Any API that accepts REST calls — which is most of them — can be called from this node. This is the node that makes n8n genuinely unlimited in scope.

The node handles authentication (API key, Bearer token, OAuth2), request methods (GET, POST, PUT, PATCH, DELETE), headers, query parameters, request body, and response parsing. For most marketing use cases you'll be POSTing to an API with a JSON body and parsing the JSON response — the node handles this cleanly with minimal configuration.

Where it becomes particularly powerful is chaining: the output of one HTTP Request node becomes the input of the next. You can call a data enrichment API (Clearbit, Apollo, Clay), pass the response to an LLM API (OpenAI, Anthropic), and then POST the generated content to a CRM or email platform — all in a single workflow.

Node 1 Workflow — AI-Assisted Content Brief Generator

Here's a workflow I use: trigger on a new keyword or topic added to a Google Sheet, call the GSC API via HTTP Request to pull the top 10 pages currently ranking for that topic, call the Anthropic API with those ranking signals plus a system prompt defining Julian's voice and content principles, return a structured content brief (target angle, H2 structure, internal link suggestions, estimated intent), and write the brief back to the sheet.

The HTTP Request node appears three times: once to call GSC, once to call the Anthropic API, once to write back to Sheets via the Sheets API. The LLM does the synthesis; the workflow handles the data plumbing. Total build time: about 2 hours. Time saved per brief: 45 minutes.

The same pattern applies to competitor content monitoring, social listening summaries, and weekly SEO performance digests — any workflow where you're pulling data from one place, passing it through an AI, and depositing the output somewhere useful.

Node 2 — The Code Node

The Code node lets you run JavaScript (or Python) inline within a workflow. This is the node that handles everything the built-in transforms can't: complex string manipulation, conditional logic with multiple branches, data reshaping, scoring algorithms, deduplication, and formatting outputs for downstream systems.

For marketers, the most common use case is data transformation. APIs don't always return data in the shape you need it — and the Code node lets you reshape it without needing an intermediate tool or a developer. You can parse a messy API response, extract specific fields, clean and format strings, run basic scoring logic, and return a tidy object ready for the next node.

The other common use case is conditional processing. If the lead score is above 70 and the company size is between 50 and 500, route to the high-touch sequence; otherwise route to the nurture sequence. This kind of logic is awkward to build with Zapier's filters; it's a few lines of JavaScript in the Code node.

Node 2 Workflow — Personalised Outbound First-Line Generator

This is one of the highest-ROI workflows I've built: trigger on a new row in a prospect list (from Apollo, LinkedIn Sales Navigator export, or a manual list), use the HTTP Request node to pull the company's recent LinkedIn posts and website content via a scraping API, pass that data to the Code node to extract the three most relevant signals for personalisation (a recent initiative, a stated challenge, a relevant product launch), then call the Anthropic API with a prompt that uses those signals to generate a personalised first line for a cold email, and write the output back to the prospect row.

The Code node does the signal extraction — parsing the scraped content, filtering for relevance, and formatting the input for the LLM in a consistent structure. Without the Code node, the LLM gets inconsistent input and the quality of the first lines is unpredictable. With it, the input is clean and the output quality is consistent.

At scale, this workflow processes 100 prospects in about 8 minutes and produces first lines that are materially more specific than anything you'd get from a generic "I saw your LinkedIn post about X" template.

Node 3 — The Wait Node (and Subworkflow)

The Wait node pauses a workflow execution for a defined period — an hour, a day, until a specific datetime. This is what turns n8n into a proper multi-step sequence tool rather than a single-trigger, single-action tool.

Combined with subworkflows (a workflow that calls another workflow), the Wait node lets you build drip sequences, follow-up chains, and conditional nurture paths that respond to behaviour. Send email 1, wait 3 days, check if the prospect opened it (via an HTTP call to your email platform's API), branch based on the result: if opened, send follow-up A; if not, send follow-up B. This is standard marketing automation logic — but built in n8n it's flexible, transparent, and doesn't require a £500/month MAP licence.

Subworkflows handle modularity. Rather than building one enormous workflow that does everything, you build component workflows (enrich a contact, generate a first line, send an email, log to CRM) and chain them together. This makes debugging much easier — you can test each component independently — and makes the overall system maintainable as it grows.

Node 3 Workflow — Lead Nurture Sequence With AI Personalisation

The full workflow: a new inbound lead lands in HubSpot (trigger via webhook). The workflow calls the enrichment subworkflow (HTTP Request to Clearbit + Code node to extract firmographic signals). Based on firmographic profile, it routes to one of three nurture paths (via a Switch node). Each path is a sequence of Wait nodes and email-send steps, with the first line of each email generated by the LLM using the enriched prospect data.

At day 7, an HTTP Request node checks the HubSpot contact for sales activity. If a meeting has been booked, the workflow terminates the sequence. If not, it continues to the next step. The Wait node makes this possible without any cron jobs, scheduled tasks, or external orchestration.

The whole system runs on n8n's cloud plan (£20/month), HubSpot's free tier, and an Anthropic API key. Total tooling cost for a nurture system that would cost £400+/month on a traditional MAP: under £50/month.

Getting Started With n8n Without Breaking Things

The mistakes I see most often: building everything in one enormous workflow (unmaintainable), not using error handling nodes (silent failures that corrupt data), and trying to automate before manually validating the logic end-to-end (you automate the wrong process).

Start with a workflow that solves one specific, contained problem — something you currently do manually that takes 20+ minutes and has a clear input and output. Build it, test it on 5 real examples, confirm the output is what you wanted, then let it run. Only then add complexity.

The compounding value of n8n is in building a library of tested component workflows that you can recombine for different use cases. That library takes time to build — but once you have 10-15 solid components, you can assemble new automation in hours rather than days.

These three nodes are the building blocks of most of the marketing automation I build — for my own pipeline and for clients. If you're building an AI-native marketing function and want help designing the automation architecture, that's work I do alongside fractional CMO engagements. Get in touch.

Related: How to Use AI for Smarter Startup Marketing · How UK Fintechs Can Actually Use AI for Growth · Services