Build an AI Email Nurture Workflow With Claude and HubSpot: From Signup to Sale on Autopilot

Here is a number that stops most marketing teams when they hear it: 80 percent of new leads never convert into a sale. Not because they are wrong prospects. Not because the product is wrong. Because nobody followed up. Another way to look at it: you are spending money to acquire leads and then ignoring four out of five of them.

Nearly two-thirds of marketing teams have no nurture program at all, according to lead nurture research from 2026. That is not a budget problem. The tools exist, the data exists, and the ROI is documented. Companies that nurture leads see 50 percent more sales-ready leads at 33 percent lower cost compared to companies that do not. Nurtured leads also spend 47 percent more. The gap is execution, not awareness.

The reason most teams do not nurture is simple: it is manual work. Writing personalised emails for every lifecycle stage, segmenting lists, scheduling sends, tracking engagement, then iterating. Doing that across hundreds or thousands of leads takes hours or weeks. But the tools that fix this are now accessible to any team with a Claude Pro subscription and a HubSpot account. This post walks through exactly how to set up an email automation workflow where Claude writes personalised emails based on HubSpot lead data, and n8n handles the routing. The setup takes about an hour, and the result runs on its own.

What you need before starting

You need three things:

1. A Claude Pro account ($20 per month) with access to Claude Desktop and the HubSpot MCP connector.
2. A HubSpot account. The free CRM tier works for basic workflows, lists, and email sequences.
3. n8n to connect everything. Free if you self-host, or $20 per month on the cloud plan.

That is roughly $40 per month total, less if you already subscribe to any of these. No API keys to manage, no custom code to write, no separate hosting.

Step 1: Connect the HubSpot MCP to Claude

HubSpot launched a native Claude connector in 2025, and by mid-2026 it has matured into a reliable MCP integration. The connector lets Claude read contact properties, deal stages, email engagement data, and company records from your HubSpot account. It can also write notes and update properties.

Step 2: Define your nurture stages in HubSpot

Before Claude can write emails, it needs to know who gets what and when. Create a static or dynamic list for each stage of your nurture sequence:

Stage 1: New lead. Someone who filled out a form or downloaded content. They get a welcome sequence that introduces your value proposition.

Stage 2: Engaged lead. Someone who has opened emails or visited pricing pages. They get deeper content and case studies.

Stage 3: Sales-ready lead. Someone with high engagement scores who has taken specific actions. They get routed to sales with a context summary.

HubSpot handles list membership automatically based on the rules you set. A contact moves from Stage 1 to Stage 2 when they open three emails or visit a pricing page. The lists update in real time, and n8n picks up the changes.

Step 3: Set up the n8n trigger workflow

n8n is the plumbing layer that connects Claude's decisions to your actual tools. When an AI agent qualifies a lead or drafts an email, n8n routes that data to HubSpot, updates lists, and triggers the send. The AI makes the creative decisions and n8n handles the execution.

The key difference between this and a standard HubSpot workflow: normally you would write every email by hand. With Claude in the loop, each email is written fresh based on what the contact has done since the last touchpoint. A contact who attended a webinar gets a different email next week than one who ignored everything.

Step 4: Write a nurture skill for Claude

This step matters more than the tool setup. Without a skill file, Claude writes generic emails. Generic gets ignored. A nurture skill is a text file where you tell Claude who you are writing to, what tone to use, and what outcomes you want.

# Nurture Email Skill

## Tone
Professional but warm. First-person company voice ("we").
Avoid jargon. Keep paragraphs to 3-4 sentences max.

## Personalisation rules
- Use the contact's industry to tailor examples
- Reference their company size when relevant
- If they have engaged with specific content, reference it
- If they have not engaged in 30+ days, use re-engagement framing

## Sequence structure
Email 1: Value intro (problem they face + how we solve it)
Email 2: Social proof (case study relevant to their industry)
Email 3: Direct offer (meeting invite or demo request)

## Banned phrases
"Just checking in", "I wanted to circle back", "Per our previous conversation"

According to practitioners using Claude Cowork email sequence skills, this file takes about 30 to 45 minutes to write the first time. Once it exists, Claude loads it automatically on every nurture task.

Step 5: Draft personalised sequences with Claude

When a contact enters a nurture list in HubSpot, n8n sends the contact details to Claude via webhook. Claude reads the contact profile, checks the nurture skill, and drafts each email in the sequence. The result is three emails personalised for that specific lead.

Personalisation at this level is why AI-driven email marketing works so well. According to Braze, AI personalisation improves click rates by 26 percent and conversions by 20 percent. Segmented campaigns generate 760 percent more revenue. If you prefer ChatGPT Work, the pattern is the same — Make.com connects HubSpot to Claude as well.

Step 6: Schedule, send, and iterate

Once Claude returns the drafted emails, n8n schedules them in HubSpot with appropriate delays: Email 1 sends immediately, Email 2 three days later if the contact did not reply, Email 3 a week after that. If the contact books a meeting during the sequence, n8n cancels the remaining emails and notifies the sales team.

After the sequence runs, you can ask Claude to analyse how it performed. Which subject lines got the most opens? Which content drove the most clicks? Claude reads the engagement data from HubSpot and gives you specific recommendations for the next iteration. Behaviour-triggered nurture emails generate up to 2.3 times higher reply rates than fixed drip sequences.

Why an email automation workflow beats manual nurture

A manual nurture sequence takes about 60 to 90 minutes to research, write, and personalise per list. With this workflow, that drops to about 10 minutes per list after the initial hour of setup. And the AI version personalises each email for the individual, not the segment. According to McKinsey data on personalisation, companies that excel at personalisation generate 40 percent more revenue from those activities.

There is another benefit that surprises most teams. The system catches up every contact who has not been followed up yet and processes them through the same sequence. That means five leads who downloaded a guide three months ago and never heard from anyone suddenly get a personalised three-email sequence. According to B2B cold email research, follow-up emails generate 42 percent of all campaign replies, yet 48 percent of sales reps never send a second message.

The same agent stack principles that apply to building AI agents apply here. Claude is the brain that decides what to say. The nurture skill is the identity layer. n8n is the tool layer that connects decisions to actions. And the HubSpot connector is the context layer that provides current lead data. For teams already automating content distribution, our content repurposing workflow uses the same Claude plus MCP approach.

Frequently asked questions

Can Claude write personalised emails based on HubSpot data?

Yes. With the HubSpot MCP connector, Claude reads contact properties, deal stages, and engagement history from your CRM. It uses that data to personalise each email with industry-specific language and timing that matches where the lead is in their buying journey.

Do I need to code to set this up?

No. Claude handles the drafting through natural language. n8n uses a visual workflow builder. The HubSpot connector is point-and-click. The most technical part is writing a Claude Skill file, which is plain markdown.

What tools do I need?

Claude Pro ($20/month), HubSpot (free tier works), and n8n (free self-hosted or $20/month cloud). Total roughly $40 per month.

Can I use this with ChatGPT instead of Claude?

Yes. n8n has nodes for OpenAI's API as well. The workflow pattern is identical. You can also use Make or Zapier instead of n8n.

How much time does this save?

Manual nurture takes 60 to 90 minutes per list. With this workflow, about 10 minutes per list after setup. AI personalisation lifts click rates by 26 percent and conversion by 20 percent according to Braze.