What Growing a Team 1 to 75 Taught Me
TL;DR Scaling a company from one person to 75 across three countries taught me that the real leadership work is emotional, not operational. The hardest lesson: you have to give away the work you're best at, usually before you feel ready. Everything I do now as a TPM at Google traces back to that.
I've been leading programs and teams for over 20 years, and if you asked me for the single experience that shaped how I lead, it wouldn't be a launch or a framework. It would be the seven years I spent building GFM Corp — growing it from just me to 75 collaborators across three countries between 2010 and 2017. We delivered 650+ websites, shipped 45+ mobile apps, and enabled 250+ small businesses a year.
Those numbers are the tidy version. The real story is messier, and it's mostly about what I got wrong before I got it right.
Growth is not a headcount problem
When you go from doing everything yourself to leading engineering, design, QA, and client management teams simultaneously, the instinct is to treat scale as an arithmetic problem. More work, more people. Hire, onboard, repeat. That's the part everyone plans for.
What nobody warned me about was the emotional tax. There's a recent conversation on Lenny's Newsletter with Molly Graham about the grief and burnout sweeping through tech right now, and she revisits her old advice that scaling a team means constantly giving away the work you love. That framing landed hard for me, because it names the thing I struggled with for years without having words for it.
When you found something, you are the best person at almost every job. You're the best engineer, the best salesperson, the best client-relationship manager, because you're the only one who has done all of it. Growth means systematically handing each of those jobs to someone who, at first, will do it worse than you would. That is genuinely painful. And if you don't get past the pain, you become the ceiling your own company hits.
The lesson I learned the slow way
Here's my honest reflection, not a war story: for a long stretch at GFM Corp, I confused being needed with being valuable. If a client escalation came in, I wanted it routed to me. If a tricky build broke, I wanted my hands on it. It felt like leadership. It was actually a bottleneck wearing a cape.
The shift came when I started building real delivery practices — establishing PSPO and Scrum across every squad, creating repeatable programs like "Idea-to-MVP" and "My First Website" so the work didn't depend on my presence. Once the system could carry the load, my job stopped being the work and started being the conditions for the work. That's the definition of leadership I still use.
The uncomfortable truth is that the moment my teams could ship without me was the moment I became a good leader — and it was also the moment I felt least essential. Both things were true at once. Getting comfortable with that tension is the whole job.
Partners taught me that trust is a program, not a vibe
After GFM Corp, I spent a stretch at Odoo as a Channel Partner Manager, working with partners across 15 countries and running 171 product demos. That period sharpened something scaling a company only hinted at: with partners, you have almost no positional authority. You can't reorganize them. You can't set their priorities. All you have is credibility and clarity.
That forced me to get precise about influence. A 71% demo conversion rate doesn't come from charisma — it comes from understanding what the person across the table actually needs and being ruthlessly honest about whether your thing solves it. When you strip away authority, leadership reduces to two things: do people believe you, and do you make their job easier? I carry that filter into every cross-functional room now.
Cross-functional leadership is the same skill, higher stakes
At Google I lead complex programs across the Android platform in the Camera and Media areas, coordinating engineering, product, UX, and partner teams. On paper it's a different world from a small agency in Panama. In practice, the muscles are identical: align people who don't report to you around a shared goal, manage risk and dependencies across multi-quarter timelines, and communicate clearly enough that nobody has to guess what "done" means.
The difference is the stakes and the surface area. A missed dependency at a 75-person agency costs a delayed website. On a platform program, the blast radius is much larger, so the disciplines around risk mitigation and escalation have to be sharper. But the core insight holds: the program manager's real product is alignment. The features are downstream of that.
Why I keep learning on purpose
Between June and September of 2026 I ran a deliberate learning sprint alongside my day job — 11 Coursera credentials, 69 courses, spanning program leadership, AI agents, and machine learning, from Google, Microsoft, IBM, Duke, Vanderbilt, DeepLearning.AI and Stanford. People sometimes ask why a Staff TPM with 20 years in would sit through 69 courses.
The honest answer is the same lesson from GFM Corp, applied to myself: the moment you decide you've arrived is the moment you become the ceiling. I don't want my experience to calcify into "the way we've always done it." I wrote more about that mindset in the future of TPMs in an AI-first world — the role isn't going away, but the version of it that survives is the one that keeps relearning its own job.
What I'm doing about this
Three intentions I'm holding myself to right now.
First, give away work faster. My default is still to grab the interesting problem. I'm trying to catch that reflex earlier and instead ask who could own it — and then actually let them own it, mistakes included. Being needed is a trap; building people who don't need me is the goal.
Second, treat trust like a program with milestones. The Odoo lesson was that credibility compounds when you're consistently useful. I want to be more deliberate about that in cross-functional settings — clearer commitments, honest status, no surprises.
Third, put the new skills to work in service of the team, not my ego. The AI-first workflows I've been building are only worth anything if they remove grunt work for the people around me, which is the whole point of how AI is reshaping TPM workflows. Leadership at scale was never about doing more myself. It was about making everyone else's work possible. That was true at 75 people, and it's true now.
Sources
Follow along
Get new posts as they publish, in whichever format you read.
