Starboy Does Not Scale

For most of my early career, work paid me back in a straight line.

I was a software engineer. Someone brought a problem, I broke it into parts, designed a solution, wrote the code, tested it, and shipped it. If I stayed after everyone had gone home, the commit history showed it. If I learnt a new framework over the weekend, Monday moved faster. If I picked up the ugly bug nobody wanted, people noticed by Friday.

The more I did, the brighter I looked. That is the starboy model. Your value is visible because your hands are on the output. I did not understand how rare that arrangement was until I lost it.

When the straight line broke

The break did not arrive when I founded a company. It came earlier, when I moved out of engineering and product into leading large teams inside a business that was scaling faster than its structure.

My output stopped being what I personally produced. The question became whether dozens of people could do work I would be proud to put my name on when I was not in the room, and keep doing it next quarter while I was somewhere else. I learned how to lead that team on the job, with guidance from my supervisor, who was the CEO. He could challenge my judgement, correct my mistakes, and help me understand what the role required.

Starting my own organisation changed the weight of that responsibility. There was no supervisor above me and no final escalation beyond me. The team, structure, standards, and consequences all ended at my desk. Whatever the organisation became was something I had to shape deliberately.

Software behaves the way you designed it. People do not. Software does not resign because nobody explained what good looked like, stop raising problems after a difficult meeting, or carry a hard year at home into a sprint review. I had spent years mastering a logical system and now owned one that was not.

Like most technical founders, I assumed building software and building people were adjacent skills. They are not even the same trade.

The mistake I almost made

The story people expect is that I held on too long and became a bottleneck. That came later. My first mistake was the opposite, and it was more dangerous because it looked like maturity.

I had read enough about founders working on the business rather than in it that I tried to graduate early. I wanted out of support tickets, difficult customer calls, and operational detail. I looked for people to hand things to before I had done those things well enough to know what a good handover required.

That is not delegation. It is abdication with a job title.

You cannot set a standard you have never reached, interview for a skill you cannot recognise, or coach work you have never done. Coaching is not encouragement. It is correction with specifics. Founders who skip the season of knowing the work in their bones often hire a title, hope for magic, and cannot tell whether what comes back is excellent, adequate, or quietly failing.

The early mess is where intuition is manufactured. Build the first version. Sit in support. Close the first partnerships. Do the reconciliation by hand once so you understand what breaks. That knowledge later lets you tell a talented person why good work is not yet good enough.

Then comes the opposite mistake. I kept reviewing designs, approving decisions, and attending meetings that ran better without me. It felt like service. It was a ceiling. Delegating before you understand the work and refusing to delegate after you understand it are the same timing error. The real skill is knowing when your role must change.

What you are doing is not good

The clearest thinking I have found on this is not in a business book. It is in Exodus 18.

Moses had become the bottleneck of a nation. Every dispute came to him, and people waited from morning until evening for one man. His father-in-law watched for a day and said something I have never been able to unhear. What you are doing is not good.

Jethro did not question Moses' ability or effort. He questioned the architecture. Select capable people. Place them over thousands, hundreds, fifties, and tens. Let them handle routine matters and escalate only the difficult cases. Teach the rules so fewer disputes arise. That is span of control, decision rights, an escalation path, a hiring standard, and documentation described millennia before anyone charged a consulting fee for it. Moses did not become less responsible. His work changed shape.

The pattern appears again in Acts 6. When growth created a problem in the daily distribution, the apostles defined a standard, selected seven respected people, and handed over the function. They protected the work only they could do without treating operations as unimportant. The text records what followed. The word spread and the number of disciples increased.

Delegation is only the first half

Hiring someone does not remove the work. It converts it. You stop producing the output and become responsible for the person and system that produce it. You have not dropped a job. You have picked up a different one.

That job has real deliverables. Define good and write it down. Hire against that standard. Give feedback people can act on. Hold difficult performance conversations. Clarify decision rights. Build a path from junior to lead. Know who takes over if someone leaves.

Your second product is the organisation's ability to make good decisions without you in the room. Nobody buys it directly, but it determines whether the product customers do buy survives.

The evidence is not soft

This is where people assume we have left engineering behind for feelings. We have not.

Nicholas Bloom and his colleagues ran a randomised field experiment with textile firms in India. The treated firms raised productivity by 17 percent in the first year and opened additional plants within three years.

Managerial time had constrained growth. Owners centralised decisions because they could not monitor delegated work, so better information reduced the need for direct supervision and made delegation safer.

The constraint was not demand alone. It was the number of decisions the owner could personally supervise.

Google found a related pattern through Project Oxygen. After studying thousands of comments, reviews, and employee surveys, coaching ranked first among eight behaviours shared by effective managers. Technical expertise ranked eighth. Technical ability still mattered, but it did not separate the strongest managers from the rest. Coaching, communication, and empowerment did.

Different industries, same lesson. Management is not a personality trait discovered after promotion. It is a set of practices that can be taught, measured, and improved.

What I do now

Five practices, in the order I would rebuild them if I started again tomorrow.

Write down who decides what. Not an org chart. A decision list showing what a lead decides alone, what needs a peer, what comes to me, and the thresholds for financial exposure in naira, KES, or dollars, as well as risk and customer impact. People escalate when guessing feels more dangerous than asking.

Define good before hiring for it. Describe excellent and mediocre output, then name the three things the person must do without help by month three. If I cannot write that page, I am not ready to hire. I am ready to hope.

Teach through decisions. AI can generate options, code, and summaries, but it does not automatically transfer the context and judgement behind a decision. Explain the customer problem, the architectural trade-offs, why the customer matters, what is urgent, what can wait, and the principles and best practices that guided the choice. People learn judgement by following the reasoning, including what later changed, not by seeing the outcome alone.

Protect the one-on-one from status updates. Status belongs in writing. The conversation is for the person's growth, the issue they are avoiding, and the feedback I owe them. Once it becomes a progress report, the space where leaders are built disappears.

Measure yourself by what happens without you. I watch which decisions reached me that should not have and how many people we promoted from inside. When I'm not available, whatever broke is my report card, not theirs.

Keep training, starting with yourself

A company rarely outgrows the habits of its founder. It mirrors them.

If the founder stops learning, the company eventually plateaus. If the founder avoids hard conversations, the culture learns that avoidance is acceptable. Your rate of learning sets the ceiling for the people hired under you.

So my reading changed to leadership, organisational design, finance, negotiation, hiring, capital, and Scripture. I went back for an MBA while running companies, not for the certificate, but because the tools of my trade had changed. I refused to operate with instincts formed in a season that had closed.

The starboy phase was not a mistake. It taught me how to build and gave me a standard worth transferring. The mistake would be treating a useful phase as a permanent operating model.

I still open an editor and use AI agents to prototype and prove ideas before delegating them. I also handle the smallest task when it needs doing instead of searching for someone else to take it.

But I no longer measure my contribution by what I personally shipped. I measure it by how many people can make a decision I would have made, and by what would still stand if I walked away tomorrow.

Products start companies. People sustain them. Institutions are what outlive us, and nobody has ever built one alone.