Every engineering leader wants a clearer read on delivery, and the means to get one have never been better. DORA and flow metrics, the signals already sitting in your tools, the trace of who reviewed what and when: it is now possible to see almost everything a team does. That is exactly the problem. The same instrumentation that can sharpen a team can, pointed slightly differently, become surveillance, and the moment it does, it stops working.
The failure is not hypothetical. When one large technology firm began capturing its employees' mouse movements, keystrokes and occasional screenshots, the stated purpose barely mattered. People experienced it as monitoring, the effort lost its legitimacy, and the backlash did more damage than any insight it produced. Measurement that feels extractive, opaque or imposed does not gather honest data. It teaches people to manage the watcher.
Measured well, the same signal builds trust. Measured badly, it builds fear. The signature insight · Measurement
Why measurement curdles into surveillance
There are two ways a good intention goes wrong. The first is the surveillance trap: the read is aimed at individuals, runs without consent or explanation, and is felt as control. Trust collapses, and with it the honesty the data depends on. The second is subtler. The moment a measure becomes a target, people optimise for the number rather than the thing it was meant to stand for. Tickets close faster and value does not move. The team gets very good at what is counted and quietly worse at what matters.
Both failures share a root. They treat the number as the point. It is not. The number is a prompt to ask why, and a number read without its context misleads more often than it informs. A slow review can be a capability gap, or a fire elsewhere, or a dependency no one owns. Only the context tells you which.
What the pattern actually looks like
- The read is aimed at people, not the system. Individual scorecards invite defensiveness and game-playing; system signals invite change.
- The metric becomes the target. Behaviour bends to the measure, and the measure stops describing reality.
- It runs in the dark. People do not know what is captured or why, so they assume the worst and act accordingly.
- A number travels without its context. The figure moves up a slide and drives a decision the why would have stopped.
What to design instead
The fix is governance, not more technology. The same signals, handled differently, become the most credible thing on the table.
- Measure the system, not the people. Read how work flows and where it stalls, the constraints that sit above any one person, not who typed what. The unit of measurement is the operating model.
- Make it transparent and developmental. People should know exactly what is read, why, and that it exists to improve the system rather than to rank them. Measurement paired with support invites engagement; measurement without it manufactures fear.
- Read every signal in context. Pair the number with the why, set a clean baseline and re-baseline against it, and never hand the organisation a single metric to optimise. Aim for readiness, the team's capacity to perform, not a scoreboard.
The takeaway
The instinct to avoid measurement because it can become surveillance is understandable, and wrong. The answer is not to measure less. It is to measure the system rather than the people, in the open, with the context attached. Done that way, measurement is the most powerful trust-builder a delivery organisation has. Done the other way, it is the fastest way to lose the room.