Once a measure becomes a target, it stops being a good measure. Construction jobs prove that every week.
Percent complete is a useful check until someone is paid, or judged, on the percentage. Then hours get booked to the wrong code, material gets received early so the number moves, and a floor that is not installed gets marked underway. The report improves. The building does not.
Milestone hit rate goes the same way. A date that was meant to tell you whether the job is real becomes a date people protect. Scope gets quietly cut so the milestone can be called done. The inspection gets booked before the work is ready, because “inspection scheduled” looks better than “inspection missed.” The metric is green. The next trade walks into a room that is not theirs yet.
Float, overtime, and safety counts rot under the same pressure. Float gets hidden in logic so the schedule still shows slack. Overtime gets pushed onto a sub so the GC’s labor report stays clean. Near misses stop getting written down once the board treats a blank sheet as a safe site. None of these numbers were bad. They became bad when hitting them mattered more than the thing they were counting.
A measure works when it can still surprise you. Planned complete against physical complete. Critical-path float. Open constraints on tomorrow’s work. Hours per unit, not hours burned. Those are good only if a bad week is allowed to look bad. The moment a bonus, a pay app, or a client meeting depends on the figure, people will optimize the figure.
The fix is not fewer metrics. It is keeping the metric on a leash. Use it to find the blocked task, the missing material, the unanswered RFI. Do not use it as the scoreboard the crew plays to. Pair every percentage with something binary and hard to dress up: inspection passed or not, units installed or not, material on site or not. If those disagree with the percentage, believe those.
Click Construction treats the number as a signal, not a grade. When percent complete, a milestone, or float drifts, the point is not a better chart. The point is the front line seeing what is off, and the fastest path to realign, before anyone has a reason to edit the truth. A measure that people can still afford to get wrong is the only measure you can trust.