PMaker home
A metric that everyone is responsible for is a metric no one isNorth StarServes as referee, not as taskComeNew activations — add an entry point, adjust a channelUseCore task completion — change onboarding, cut stepsStayWeek-1 retention — improve re-engagement, build habitsReturnActivity after returning — personalized content, recommendationsEvery path gets a metric, a set of actions, and an owner

The North Star is a direction, not a task. Split it until every path can be acted on, measured, and owned separately — then it stops being a slogan and becomes daily work.

Splitting the North Star

A North Star only tells you whether the whole is getting better, not what to do. Split it into paths you can act on separately, and it stops being a slogan and becomes daily work.

A North Star can only tell you "whether the whole is getting better" — not "what to do." Split it into paths you can act on separately, down to where every path has an owner and a number to watch, and the North Star stops being a slogan and becomes daily work. This is the core of the North Star framework: the metric is the outcome; the inputs are what teams can actually push on day to day.amp-framework

What you'll run into:

  • The North Star is set, but nobody knows what to do this month to move it
  • Everyone is "responsible for the North Star," which means no one is
  • Only the top-line number exists, no intermediate metrics, so you can't tell which link deserves credit

Why one metric isn't enough

The North Star works as a "referee," not as a "task" — it answers "did this quarter go well," not "what do we do this week."

Before splitting, make sure the North Star is real. If the number at the top is wrong, the paths just help you push harder in the wrong direction.

  • The top-line metric is a result, not an action. "Weekly active users" won't tell you what to do next. A number you can only look at after the fact can't be a manual in advance.
  • Results can only be reviewed; intermediate metrics are the handles. Went up — more new users, or fewer churned? Only a split can answer that. Split apart, every link has something to do.
  • Unsplit, nobody claims responsibility. A metric everyone is responsible for is a metric no one is. Split into paths, and each gets a concrete owner.

How to split it

Split along the user's path — come, use, stay, return — and land one independently actionable metric per step:

Link Metric for this step What can move it Who watches it
Come New activations Add an entry point, adjust a channel Growth
Use Core task completion rate Change onboarding, cut steps Product
Stay Week-1 retention Improve re-engagement, build habits User operations
Return Activity after returning Personalized content, recommendations Algorithms

At this layer, "what to do this month" has an answer: every path has a metric, a set of actions, and an owner.

  • Draw the user path that leads to the North Star. From come to use, stay to return — draw it as a walkable path. Without the path, the split is guesswork.
  • Land one independently visible metric per step. Come looks at channels, use at completion, stay at month-1 retention, return at re-engagement. Don't share one number across several paths.
  • Every path must be movable by one concrete action. "Add an entry point," "change an onboarding step," "tune a reminder" — if you can't move it with one action, it's not a path, it's something you just watch.
  • Every path needs its own instrumentation. Without tracking, the path is just a story. Ship the instrumentation together with the feature.

Did you split it right

Run the split through three checks:

  • Every path can be moved independently. One action should touch only this path and nothing else — that's what makes targeted work possible.
  • Every path has data to watch. A path that isn't tracked and can't be computed isn't a path. Confirm the data source together with the path.
  • Every path has an owner. Owned paths move forward; unowned paths only ever appear in the weekly report.

And don't forget the upstream: no matter how finely the paths are split, a wrong North Star makes it all moot. When all three checks pass, the split is done — then revisit it quarterly: have the paths changed, have the owners changed, does the metric still respond.

References

  1. About the North Star Framework — Amplitude