GTM Atlas — Retention is a culture problem

Author: Josh Epstein, CRO at Coder
Source: atlas.attio.com/retention-is-a-culture-problem
Date: May 6, 2026

Who is Josh Epstein

CRO at Coder (AI development infrastructure). Drove 5x revenue growth in 20 months. Before that, six years at HashiCorp building the Global Accounts business from zero to $175M+ through their IPO.

Core Thesis

Retention is the culture you build on day one. If your team's first instinct is to close the deal and worry about the customer later, that culture permeates and you can't pull it back. If you're not planning for expansion before the first deal closes, you're not planning to win.

Key Frameworks

The Second Deal Is the Signal

You always close the first deal — impressive demo, friendly referral, rapport you built. There are a million reasons the first one lands.

The second deal is the signal that customers are getting value and have some confidence that you will be there to support them.

The 7 over 3 pattern: On average, a customer is seven times more valuable after three years. A $100K deal becomes a $700K customer. Tell customers that openly. The math runs the other way for them too: they get 20-40x that value back. They should be happy to pay because they know it's scaling the business you're building for them.

The Third Founder

Most early companies have two founders: engineering/product-led and GTM-led. Usually the GTM lead is an engineer who's a little more personable.

The real hire comes later: a program manager, chief of staff — the person who can translate between customers and the people building the product. When the founder paints an 18-month vision, the third founder is the one who says: here's what I can ship in 90 days, but you have to invest (buy software and provide feedback). We're serving the food while we're cooking it.

Fight the urge to put your smartest people in front of customers. The conversations belong to people with customer empathy who work to define the value customers get from your solution.

The 70/80/90 Metric

The single best predictor of customer retention and expansion:
- 70% of licenses deployed
- across 80% of customers
- within 90 days

Speed is the biggest single signal about whether a customer will renew and expand.

Correlation runs backward too. Customers who went through formal technical validation (success plan, CSM attached, engaged from the start) hit 70/80/90 a hundred times out of a hundred. Customers who skipped that loop hit it 40% of the time.

Customers underscope constantly — buy licenses for a team of 300 when they should buy for 100. They hit 70% deployment in 30 days and have to scramble to expand. We told them to buy bigger; they didn't believe us.

Churn Is Visible 180 Days in Advance

  • They stop opening tickets
  • They stop returning CSM calls
  • Third independent support ticket means the tool has spread across teams
  • Knowledge-based self-serve means someone's building an internal practice

Old line: if a company has an email distribution list with the word Coder in it, we win.

Think Two Years Out

Force everyone to think farther out than they are comfortable with:
- CEO: 2-3 years ahead
- Every executive: 1-2 years minimum
- Teams and leaders: 3-6 quarters

Set up customer communication channels before you need them: customer journey mapping, reference architectures, a way for customers to raise their hand to give feedback, monthly newsletters, documentation of why every deal closed, a public knowledge base.

Act bigger than you are. Those automated processes give customers confidence they're dealing with a bigger company.