You Can't Measure a Feeling

You want to renovate your kitchen. You agree on a budget and describe what you want (“open it up, make it brighter!"). Eight months later, the kitchen is finished to a high standard and looks beautiful – but it’s just not what you imagined it would be. The island is in the wrong place, there’s only one cupboard, and is that a toilet!?!? The work isn’t bad. The contractor built exactly what was asked for. The problem is that the ask was never clearly specified.

That gap between a vision and a target is where most transformation measurement quietly fails, long before anyone creates a dashboard.

The wrong question

Ask most leaders how they'll know a transformation worked, and the answer arrives fast: satisfaction scores, adoption rates, engagement surveys. These aren't bad instruments, but they're answering a different question than the one that matters. A satisfaction score tells you how people felt about an experience. It says almost nothing about what they did differently afterward, which is the entire point of a transformation. Everything else is scaffolding.

Sentiment survives this vagueness because it's cheap and already built into most tools. Behavior doesn't, because measuring it requires someone to have decided, in advance, exactly what it was supposed to look like. That's the actual failure, not the wrong KPI, but the absence of a defined target to measure against at all.

Vision isn't a target

"Improve collaboration" and "modernize how people work" are legitimate aspirations, but neither describes an observable action, so neither can be measured. You can't check whether "improve collaboration" happened. You can check whether cross-team requests that used to take a week of email now get a same-day response in a shared channel. One is a direction. The other is a target.

This is the renovation's failure again. "Open it up, brighter" was a direction. You only get the kitchen you want if you do something boring and unglamorous first: write down what "done" looks like (island placement, storage required, number of toilets) and agreed on it with the contractor as the thing both sides would check the finished work against. Transformation efforts skip this constantly, because naming a specific behavior in advance is a commitment, and it's easier to gesture at a future state and sort the measurement out later.

Why "later" doesn't work

A baseline defined after the fact isn't a baseline, it's a narrative. Decide what "success" looks like once you already sense how things are landing, and you'll gravitate, without any bad faith involved, toward a definition the data can support. The only way a measurement can tell you something you didn't already believe is if it was fixed before the outcome was known.

This is where well-run transformation work quietly loses its diagnostic value: not wrong metrics, but metrics chosen with the benefit of hindsight, which means they've stopped being able to surprise anyone. The fix is a discipline, not a technique. Define the specific behaviors the change should produce, in observable terms, before the work starts, and treat that definition as fixed.

It's the same instinct behind a good diagnostic assessment done up front: establishing what you're checking against, independent of how things turn out.

What this buys you

Sentiment drifts slowly and reports even more slowly - a quarterly survey tells you, three months late, that something felt off. Specific behavior, tracked from week one, shows divergence almost immediately. If the target was "requests get resolved in the new system without falling back to the old workaround," you'll know within weeks whether people have quietly built a parallel process to avoid it.

That's the real difference between a scorecard and an early-warning system. A scorecard tells you, after the fact, how the story ended. An early-warning system tells you, while there's still time to act, whether the thing you actually meant is happening at all.

Most transformation work never gets that signal, not because the behavior was unmeasurable, but because nobody named it early enough for the measurement to mean anything.