The engineering leader’s real job is decision throughput
Teams are rarely blocked on capacity. They are blocked on decisions nobody is empowered to make, waiting in a queue nobody is measuring.
Ask an engineering leader what limits their team and you will usually hear about headcount. Watch the same team for a fortnight and a different picture emerges. Work is not stalled for lack of hands. It is stalled waiting for a decision — about an approach, a dependency, a priority, a trade-off nobody has been given the authority to settle.
Capacity is the constraint everyone can see because it appears on a budget. Decision latency is invisible, unmeasured, and in most organisations the larger of the two.
Where decisions go to wait
They wait for a meeting that recurs fortnightly. They wait for someone senior who is travelling. They wait because three people each believe another owns it. They wait because the person who could decide has learned that deciding wrongly is punished more reliably than not deciding at all.
That last one is the most corrosive and the least discussed. A team that has watched a colleague criticised for a reasonable call made with incomplete information will escalate everything thereafter, and the escalation is entirely rational.
Improving throughput
- Name the decider for every significant open question, in writing. Ambiguous ownership is the single largest source of delay.
- Push decisions to the lowest level that holds the relevant context, which is usually lower than the level currently holding them.
- Distinguish reversible from irreversible. Reversible decisions should be made quickly by whoever is closest; irreversible ones deserve the process.
- Set a default. “If we have not decided by Thursday we do X” converts an open question into a deadline and removes the incentive to stall.
- Defend people who make reasonable calls that turn out badly. The cost of a wrong decision is bounded; the cost of a team that has stopped deciding is not.
What to measure
Most engineering metrics describe output. Very few describe the conditions that produce it. The more diagnostic questions are simpler: how long does a decision sit before it is made? How many open questions are currently blocking work? What proportion get escalated past the person who had the context?
Leaders who move those numbers usually find that the headcount conversation becomes less urgent, because the team they already have starts finishing things.
Let's build the system your business will run on next.
Tell us where it hurts. We'll bring the architects, engineers, and delivery model to fix it — and scale it.