In my last month leading the infrastructure organization I'd built over several years, I wrote a transition plan. It named an interim leader, assigned owners to 12 priority initiatives, documented decision rights, and set a schedule for handoffs. It was a good plan for keeping the work moving.
It named the people responsible for that work. It said little about what those years had meant to them, what they had learned, or how their judgment had developed. A transition plan has a different purpose, but writing mine made that gap difficult to ignore.
That month, I came across a clip of Clayton Christensen's 2012 TEDxBoston talk, How Will You Measure Your Life?. He was asking how to measure a life. My transition plan raised a narrower question: what evidence did I have of my impact as a leader?
I had to run the organization on rollups. The people in it experienced me one conversation at a time.
Leaders need the rollup
Christensen starts with a practical constraint, which is that finite attention forces us to aggregate information. We need summaries to understand an organization, but those summaries cannot establish our impact on individual people. In his religious framing, God has no such limitation and can consider each person's life without reducing it to a total.
The leadership question matters regardless of whether we share that belief. Operational metrics are necessary evidence of performance. They are incomplete evidence of what people experienced, learned, or became capable of under our leadership.
The organization I led had over 250 employees and contractors. I had seven direct reports, and management reporting helped me understand work I couldn't observe directly. For senior leadership, I built a value slide on our cloud modernization program. It had four measures: total estate availability, unit cost economics, cycle time reduction, and optimized labor mix.
Optimized labor mix describes staffing in terms of an economic result. It says little about what a change means for the people doing the work. I chose it anyway. Senior leaders needed a concise account of performance, and I needed to provide one.
When I drafted the job description for my successor, I asked for a leader with a metrics-driven focus on outcomes. I'd write that line again. Availability and cost matter to the teams that depend on a platform, and individual stories cannot replace a consistent view of its performance.
But an attrition rate can't tell me which engineer has stopped learning. A green availability chart doesn't show whose weekends helped keep it green. A measure can be accurate while leaving out an experience I need to understand.
The result doesn't show the judgment
During a late-night review of a Kubernetes network component upgrade, an engineer raised her hand and recommended rolling back. She wasn't confident we could resolve the issue within our change window.
We had discussed this discipline: follow the plan, respect the window, and be prepared to roll back and try another day. I know the temptation to believe a fix is one more minute away. I have caused enough production incidents to understand what that confidence can cost.
What stayed with me was that she raised the concern herself. Under the kind of pressure that usually makes discipline give way, she applied a principle we had discussed.
A rollup might record a completed upgrade or a rollback. What I remember is an engineer recognizing uncertainty, speaking up, and exercising her own judgment about when to stop.
Participation is easier to count than development
My own development goal had a similar limitation. In July, I wrote a performance goal to build the leadership bench and a development culture across the organization. I measured it through succession planning, development goal coverage for everyone, and my four mentees.
Each measure recorded something useful, but none could tell me on its own whether people had developed. Goal coverage could reach 100% while every goal remained untouched after someone typed it in. A succession plan could name a person without showing whether they had received the experience needed for the role.
Even the four mentoring relationships I tracked were just a count. The harder question was what our work together had helped each person do.
This is a different problem from losing detail in an average. The measure itself stops at participation. Adding more detail about meetings or completed forms would still leave the outcome unanswered.
The answer needs to include what the person wants. Development might mean deeper technical expertise, better judgment, greater confidence, or responsibility they can sustain. A promotion is one possible outcome, and the leader shouldn't assume it is everyone's destination.
Attention follows the fastest evidence
Later in that talk, Christensen describes how our actual priorities emerge from the way we allocate time and energy. An extra half-hour tends to go toward whatever produces the most immediate evidence of achievement. Over time, those choices can produce a life we never intended.
I recognize the same pressure in engineering. A feature with a deadline and a sponsor has an advantage over reliability work whose benefit is an incident that never happens. Without explicit priority and capacity, that work can lose one planning cycle at a time. Each decision can look reasonable in isolation.
A leader's attention can follow the same pattern. A dashboard offers a visible update every morning. Developing someone's judgment is harder to assess in a weekly review, and the conversation about it can feel easier to postpone.
I've written in Work Is Hard Enough about how much of leadership happens through people. That includes urgent decisions, conflict, and coordination. A week filled with those conversations can still leave little room for someone's development.
Protecting development time matters, but a full calendar cannot answer the question either. A recurring one-on-one can become another status meeting. Mentoring can continue for months without addressing what the person needs.
Reserve the time, then examine what changed
My public user guide includes commitments to clear roadblocks, advocate for the team, and share credit while owning blame. Those commitments need to become decisions and actions that someone else can describe.
Reserving time before the quarter fills up—one-on-ones, skip-level conversations, and mentoring—is only the baseline. What matters is examining the opportunities around that time. Does the person have meaningful work to practice on, clear authority, useful feedback, and support when a decision is difficult?
For a leader responsible for hundreds of people, this has to extend through the managers in the organization. A handful of personal mentoring relationships cannot establish whether everyone else has access to development.
Alongside goal coverage, I'd want managers to explain what someone can now handle, what experience helped, and what support they still need. I'd then use skip-level conversations to understand how employees describe those same opportunities. The point is to check whether our account matches their experience and act where it doesn't.
That requires more than collecting success stories. Consistent reporting can help counter the tendency to notice whoever is closest or escalates most visibly. Individual conversations can expose what that reporting misses. Both are necessary, and both can mislead when we treat them as sufficient.
My own interpretation needs scrutiny too. I can believe I helped someone grow when their progress came mainly from colleagues, experience, or their own persistence. The check I trust is concrete. Can I describe what changed and explain my contribution, and would they tell the story the way I do?
Try that with three people outside your usual circle of visible performers and frequent escalators. Write down what each wanted to develop, what opportunity they received, and what you observed. Then ask how they experienced it, including what remains missing. A difference between the accounts gives you something specific to address.
My transition plan had an owner for each of 12 initiatives and a date for every handoff. Those details mattered to the person taking over. They couldn't tell the whole story of the people doing the work or what my leadership had contributed. What people had learned while we worked together needed no handoff, because it already belonged to them.
Related reading
