Six groups, each with the honest response and the shortcut side by side, and a way to mark the ones that would happen where you work.
The problem
Every metric lands on people, and people are good at metrics. The question is never whether the number will move — it is what they will do on the way, and what quietly gets worse while it goes up.
That conversation almost never happens before the metric is set, because the people who would have the answer are the ones the metric is about to land on.
What it does
A metric
Defined to the edges, because the edges are where the shortcut lives.
Its people
Every group it lands on, not just the team that owns it.
Two versions
What each group does if they are being honest, and what they do if they just need the number.
Guardrails
A counter-metric for each shortcut you recognize, and how long before it shows.
What it showed
Groups the number lands on
6
Including the ones who do not own it and will still change what they do.
Guardrails it produced
5
One counter-metric per recognized shortcut, with the lag before it would show.
Reading the definition back is what does the work. “The count of accounts that open something inside seven days” means the cheapest way to succeed is to get more accounts to open something — not to help anybody — because opening a reminder email counts as activity, and activity can be manufactured by sending more email.
Nothing in that definition requires the person to have wanted it. That is not a prediction about bad actors; it is a reading of what the number literally rewards, and it is available before anybody is measured on it.
What it cannot do
It maps the shortcuts you can think of. The one that costs you is usually the one nobody in the room had a reason to imagine.
Marking a row “already happening” is a judgment, not a measurement. The guardrail it produces is what turns it into one.
The long version of the argument behind it is The 2026 product leader’s toolset.