Founder's GTM

Founder's GTM

Claude Code GTM BUILD #1: EMAIL Infrastructure

A marketing agency asked me to restart their cold email in one of the most spammed markets around. Part 1: the setup, from “yes” to the first send, with every prompt.

Jasper Vanu | Claude for GTM's avatar
Jasper Vanu | Claude for GTM
Oct 20, 2026
∙ Paid

I almost said no to this one.

Last summer, the founder of a marketing agency reached out wanting to revamp their GTM with cold email. Their buyers: Shopify brands.

Two things made me hesitate:

  • I don’t usually work with ecommerce at this scale.

  • Shopify brands get pitched constantly. Every agency sends them the same “free email audit”. Their inboxes are some of the most spammed in B2B

But they really wanted to do this. So I said yes, without high expectations.

I’d use everything I’ve learned from over 10 years of outbound to give it a fair shot.

This is part 1 of a live build.

I’ll split the project into phases and publish a new part at each one, with the numbers, good or bad.

Here’s one number to start with: the first meeting came 15 days after the first email went out.

That story comes in part 5. First, the foundation everything else runs on.

Free when you subscribe: the GTM Starter Kit. + Claude skills that find your next buyers and write the emails that book the call.

🗺️ The build in 7 phases

  • Phase 1: EMAIL INFRASTRUCTURE

  • Phase 2: THE MARKET AND THE LIST

  • Phase 3: THE SIGNAL

  • Phase 4: THE COPY AND THE PAGES

  • Phase 5: LAUNCH AND FIRST SIGNALS

  • Phase 6: RUN AND FIX

  • Phase 7: RESULTS AND THE VERDICT

This part covers phase 1. Follow along, and you can build your own system the same way.

🏢 The client

A remote ecommerce marketing agency.

Around 60 people, seven figures in revenue, selling retention services to Shopify brands.

By 2025, every new client came from inbound or referrals. Cold email was switched off, and for good reason:

  • Two years of sending it themselves: about 390,000 people emailed, and the cost per attended call climbed from $324 to $905

  • A pay-per-meeting agency: 26 meetings, 0 clients

When the founder and I first spoke, he was clear:

“I don’t want an agency.”

“I just want to avoid very expensive tests.”

🧰 The stack for this phase

  • Planning: a Claude project, with a separate chat for infrastructure

  • Domains and mailboxes: Primeforge

  • Inboxes: Google Workspace and Microsoft 365

  • Sending: Smartlead

  • Claude Code: connected to Primeforge and Smartlead, to check the whole setup.

📍 Phase 1: Email infrastructure

Seven steps, from “how much will we send?” to the first email out:

  1. Set the volume first

  2. Plan the setup with Claude

  3. Buy domains and mailboxes on day one, with Primeforge

  4. Set up DNS on every domain

  5. Warm up, and start slow

  6. Set up your sending platform, in Smartlead

  7. Load your suppression list before anything sends

Below: every step in full, with what we did, what I’d do differently, and the prompts.

Inside:

  • The 2 questions that decide how big your setup needs to be

  • The infrastructure prompt that planned our domains, DNS and warmup

  • Why we mixed Google and Microsoft inboxes

  • The exact Smartlead settings we launched with

  • The suppression mistake that hides old leads without any error

Phase 1, step by step

Step 1: Set the volume first

Every number in your setup comes from one decision: how many emails a month you’ll send. So settle that before you buy anything.

We started with two questions.

Why did it fail last time? I asked the agency to write down what they sent, to whom, and what came back. Their answer: a commodity offer, unfiltered lists, nobody keeping score, and sending in bursts right as Google and Yahoo tightened their rules.

That last one is an infrastructure problem. So this time, the rule was steady daily volume, never bursts.

What are we promising? We agreed on 8,000 emails a month. That number sized everything in steps 2 to 5.

Restarting a channel someone else ran? Run this with them first:

I'm restarting cold email for a company that stopped it.
Interview me about the last time they ran it, one question
at a time:
- what they sent, to whom, and how many
- what came back: replies, calls, clients, cost per call
- what they changed along the way, and what stopped

Then tell me the 3 most likely reasons it failed, ranked,
with the number behind each one. Flag anything you're
guessing. End with what we must do differently this time.

Step 2: Plan the setup with Claude

Every decision ran through one Claude project, with a separate chat for infrastructure. The project instructions set how Claude works with me:

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2026 Jasper Vanuytrecht · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for writing, podcasting, and media founders.