A wrong metric systematically misleads the organization
Aliases: false north star · metric misdirection · systematic misdirection
What it is
Once the north star is the wrong count, the error is amplified by planning, hiring, the experiment queue, and external narrative; it does not stay on a report. That is wrong north star misdirection. If daily opens are treated as the direction of a content product, the organization systematically rewards designs that reclaim attention rather than designs that finish reading or making. After a few quarters, team structure, ranking strategy, and the definition of success have grown around the error. Choosing wrong is not the same as being gamed later: the direction already points the wrong way before anyone exploits a loophole.
Why it happens
The aligning force of a single direction does not switch off when the choice is wrong. Hiring screens for people who can move that number; the experiment queue ranks work expected to move it; failure is defined as “that number did not move,” even if user value has already changed. The misdirection is systemic because it changes what counts as progress, not one judgment. Typical mistakes include treating platform scale as user value (registrations), treating a middle state as completion (opens rather than done), and treating a company revenue moment as experience success (quarter GMV). Once those counts set direction, contrary evidence struggles to enter: task failure, regret, and the departure of high-quality users will not veto the roadmap unless a counter-count exists. Correction gets costlier with time, because people and systems have already coupled to the wrong direction.
Studying it
Run a counterfactual on candidates: if the last four quarters had optimized only this count, which projects would have shipped and which would have been cut, then compare that mix with independent evidence of user value. Test error-direction sensitivity: list levers the count would most readily reward that are unacceptable on value, and see whether those levers already appear under the current incentive. Comparing adjacent products or historical stages also works: when direction changes, does the approved experiment mix shift as a block. Evidence of misdirection is a structural shift in the action mix, not whether a single number looks good.
Where it stops holding
Direction should change with stage; that does not make every change proof the previous star was wrong. Expiry and error need separate write-ups. A wrong local KPI can often be offset by another group; a wrong north star has no offset. Sometimes the error is the window: a count that is right for a year misleads monthly decisions—that is a cycle problem, not a wrong construct. Using a composite as the north star turns “wrong” into an unauditable weighting error, which is harder to catch.
Applying it
- Write three levers the current north star would systematically reward that the product cannot accept; if those levers are already on the roadmap, pause that direction.
- Each quarter, check the direction against value evidence that is not in the bonus (task completion, repeat use, unprompted recommendation); persistent divergence starts a reselection.
- Reselection must rewrite experiment gates and planning templates, or the old direction will keep ranking work by inertia.
- Stamp the start date and intended stage on the north star, so a count that was right for one stage is not extended into the next.