How to define and evolve your ICP to scale from zero to ten million

Executive overview

Most founders fail to scale because they try to sell to everyone. A tight ideal customer profile (ICP) — defined by firmographic attributes, not pipeline signals — is the single most predictable lever for early growth.

The green/yellow/red framework forces explicit choices: who to pursue, who to nurture, and who to walk away from. As you scale, 90% of resources go into the proven ICP; 10% funds experiments to discover the next one.

The ICP is who you're exploiting today — not your total addressable market.

Defining ICP correctly

• ICP is 5–10 firmographic attributes (revenue, employees, location, org structure) — things a rep can verify before the first call
• Avoid defining ICP by inbound demand or lowest CAC; optimise for maximum lifetime value (LTV)
• Green = cold-call and actively pursue; Yellow = sell only if inbound engagement is high; Red = do not sell, put into multi-year nurture
• ICP is owned by the founder and product leader — not the sales team

ICP at product-market fit

• Start with a hypothesis (e.g. healthcare/finance, 50–500 employees, North America)
• Deliberately place 30–40% of early conversations in the periphery to pressure-test the hypothesis
• When periphery customers fail to succeed with the product while core ones thrive, you have conviction
• Pivot until the bulls-eye is confirmed; only then constrain the funnel

ICP at go-to-market fit

• Once conviction is established, stop exploring — fill the funnel exclusively with green ICP
• The most predictable path from $0 to $1M to $5M is strict ICP discipline
• Investors want large TAM; manage this by showing a credible expansion roadmap, not by loosening ICP prematurely

Expanding ICP without derailing growth

• A common mistake: committing to new markets (Europe, enterprise) in Q4 without prior experimentation
• Treat expansion exactly like a new product-market fit cycle — it takes time
• Allocate 90% of resources to the known ICP; reserve 10% to run 2–4 expansion experiments in parallel
• Place experiments 1–2 years before you need them to produce revenue
• Expansion can be a new geography, a new segment, a new channel, or a new technology (e.g. AI-augmented SDR)

Who to put on expansion experiments

• Not your top closer — they need a defined playbook that doesn't exist yet
• The right profile mirrors your first-ever sales hire: adapts to change, does consultative selling, works closely with product and engineering
• Expansion experiments are a strong career growth path for the right rep