The Promise and Pitfalls of People-Performance Management Metrics

People-performance metrics have become an industry unto themselves. Entire companies exist to gather, analyze, and report on worker activities. They sell dashboards full of charts and ratios that claim to reveal who's performing, who's slacking, and who deserves recognition. The problem is simple: most of it is noise. In organizations built on thought-work, especially in software engineering, the usual ways of measuring performance are close to worthless. They produce more noise than insight, and they incentivize what's become known as "performance theater."

Lines of code written, story points completed, tickets closed, or even abstract notions like "impact" are all highly contingent on logistics. Who happened to get staffed on the high-visibility high-impact project? Who was the subject matter expert assigned to the low-visibility low-impact work (because they were the only person who could do it)? Who went on holiday during the quarter (and therefore couldn't contribute as much output)? None of these metrics tell you whether someone is actually good at their job. They're just proxies, and proxies are slippery. They give a false sense of objectivity while masking the reality that most performance assessments are still a mix of politics, perception, and circumstance.

If the agreed work was delivered, and it met the explicit terms, then the individual performed. If not, then they didn't. Everything else is noise.

There's a cleaner, fairer way to approach this: treat performance like a contract. A commitment is made. A commitment is fulfilled, or it isn't. That's the heart of evaluation. If the agreed work was delivered, and it met the explicit terms, then the individual performed. If not, then they didn't. Everything else is noise. This model forces leaders to do their jobs upfront: set clear expectations, scope commitments correctly, and avoid moving the goalposts after the fact. And it protects workers by ensuring they're judged on what was actually agreed, not on opinions or optics.

This approach also respects the natural variability in human capacity without lowering standards. Some people can take on more; some need tighter scoping. The leader's role is to set commitments that stretch without breaking, to prevent what I call "cognitive overdraft." The worker's role is to fulfill those commitments, nothing more and nothing less. In this way, performance management becomes fair, consistent, and rooted in meritocracy.

Think about how we hire contractors. If I bring someone in to renovate my kitchen, the measure of performance isn't how many nails they hammered or whether I felt entertained by their presence on-site. It's whether they delivered what we agreed to, at the agreed quality, within the agreed terms. The same standard should apply in the workplace. If a commitment is sufficiently important, put it in writing. If it's not, it doesn't count.

Aside: On "business impact." Measuring performance by post-hoc impact is one of the worst practices out there. It punishes workers for the company's own missteps — if leadership asked them to build the wrong thing, that failure belongs to leadership, not the people who executed.

The promise of people-performance metrics is seductive: objectivity, precision, and insight at scale. But the reality is that these systems too often produce theater instead of truth. The way out isn't more dashboards — it's more clarity, more fairness, and more discipline around commitments. Stop measuring the noise, and start measuring what matters.