What I'd do differently building a B2B GTM team from scratch
I've built GTM teams across multiple ventures. Every one had the same failure mode: the team couldn't run without me in the middle. Here's what I'd do differently — and the three things I'd never skip again.
If I had to start over building a B2B GTM team, I'd do three things differently from day one. I'd define the ICP from closed-won data before spending a dollar on outbound. I'd build a system that runs without me prompting it, instead of relying on my own consistency. And I'd treat knowledge capture as infrastructure, not documentation — because everything I built fell apart when people left and took the context with them.
What I got wrong about ICP in early ventures
I defined the ICP in a meeting. It looked right on a whiteboard: company size, industry, job title. What I never checked was whether those criteria matched who actually closed. When I finally pulled the data on our best customers, the profile was 40% narrower than what we'd been targeting. We'd been sending outreach to companies that fit the description but not the pattern. The fix is mechanical but requires discipline: pull 5–10 closed-won accounts, identify what they have in common beyond the obvious, and set that as the ICP. Not the ICP you want. The ICP that buys.
What I got wrong about consistency
I treated outbound like a project. Block a week, run a campaign, get some pipeline, move on. The problem: outbound is a motion, not a campaign. Pipeline you build in week 1 closes in week 10. If week 2 through 9 are empty because something more urgent came up, you've lost the compounding. Every time I stopped outbound because something needed my attention, the pipeline hole showed up two months later with no warning. The fix isn't more discipline. It's removing yourself from the critical path of execution.
What I got wrong about knowledge
Every time a strong SDR left, we lost 6 months of learning. What segments replied, what objections mapped to what company types, what message angle worked for technical buyers vs business buyers — all of it lived in their head and in their call notes that nobody read after they left. I tried playbooks. They went stale within 60 days. The real fix is a memory architecture that updates automatically from every cycle's outcomes and persists regardless of who's on the team. That's what I built Ektie's three-tier Brain to do — not because it's a clever product feature, but because I'd lived the failure mode too many times.
What I'd prioritise differently from day one
Week 1: define ICP from data, not assumption. Month 1: get outbound running in a system that executes without daily founder input. Month 3: establish a supervision loop where someone (or something) reviews every cycle and writes what worked into persistent memory. These three things, in this order. Everything else — better messaging, more channels, more sophisticated targeting — compounds on top of this foundation. Without it, you're rebuilding from scratch every time the team changes.