Where this sits
Session a1 gave you the driver-tree mindset: a top-line number is a lagging outcome, and behind it sits a tree of levers. Today we sort those levers by when they move. A metric near the root moves late and reports the score; a metric near the leaves moves early and warns you it is coming. Get this right and your operating reviews change character - you stop reacting to last quarter and start steering the next one. Get it wrong and you are always managing by the rear-view mirror.
The two kinds of indicator 8 min live
A lagging indicator is an outcome - revenue, ARR, market share. It moves late, it is what the board sees, and by the time it has moved the cause is weeks in the past. A leading indicator is an input - sign-ups, pipeline created, traffic, trials started. It moves early, it is what a team works on day to day, and it warns you that the outcome is about to move. The Amplitude North Star framing puts one lagging output at the top and a handful of leading inputs beneath it, on exactly this logic: inputs move first.
LiveThe rear-view mirror and the road ahead3 min▶
Lagging metrics are honest but late. By the time revenue confirms a problem, the sign-ups that would have become that revenue stopped weeks ago - so the confirmation arrives exactly when it is least useful. Leading metrics are the opposite: earlier, noisier, and steerable. The North Star Playbook is built on this: choose one lagging output that captures the value you deliver, then a small set of leading inputs that a team can push and that reliably pull the output along behind them.
- Lagging = outcome. Revenue, ARR, market share, retention. Late, board-facing, slow to move.
- Leading = input. Sign-ups, pipeline, traffic, trials. Early, team-facing, quick to move and quick to warn.
- Same tree, sorted by time. Roots lag, leaves lead. The distinction is not a new tree - it is reading the one you have by when each level moves.
The team that only watched revenue. A sales org reviewed one number every week: closed revenue. It looked fine right up until it did not, and then every quarter ended in the same surprise. The fix was not more discipline on revenue - it was watching pipeline created and stage-conversion, the leading inputs that had been quietly sliding for six weeks before revenue ever flinched. They had been reading the scoreboard and calling it forecasting.
Self-studyNorth Star output and its inputs2 min read▶
Amplitude's framing is worth internalising. The North Star is a single lagging output metric that expresses the value customers get - not revenue directly, but the thing that produces it (for a marketplace, say, weekly transacting users). Beneath it sit three to five input metrics that teams can move, chosen because moving them reliably moves the North Star. The output keeps everyone aligned on the same destination; the inputs are where the work actually happens.
- One output. Lagging, singular, the shared destination. Too many North Stars and you have none.
- A few inputs. Leading, controllable, each owned. The inputs are the steering wheel.
- The link must be real. An input earns its place only if moving it actually moves the output.
Steer by leading, report by lagging 7 min live
Here is the rule that reorganises how you run a team: you manage the leaves and you report the root. The deeper into the tree a metric sits, the earlier it warns - so operating conversations should live near the leaves, where you can still change the outcome, while the board conversation lives at the root, where you account for it. A leading metric only earns that role if it is both predictive (it actually forecasts the outcome) and controllable (a team can move it). Miss either and it is just noise on a chart.
LiveThe predictive-and-controllable test3 min▶
A leading metric that is predictive but not controllable - like a macro index that forecasts your demand but that you cannot touch - is useful context, not a lever. A metric that is controllable but not predictive - one a team can move all day without the outcome ever budging - is a vanity dial, and worse, an invitation to game. A metric worth steering by has to be both: pull it, and the outcome reliably follows. That double test is what separates a real leading indicator from a number that merely looks busy.
- Predictive: when this moves, the outcome moves after it, reliably enough to trust.
- Controllable: a named team can actually move it with the work they do.
- Both, or it is out. Predictive-only is context; controllable-only is theatre. Steer only by metrics that pass both.
Self-studyManage the leaves, report the root2 min read▶
This is why the two audiences of a metric tree are different. Teams live at the leaves: they have weekly targets on pipeline, traffic, activation - the inputs they can push this week. The board lives at the root: it wants the outcome, the trend, the story. A well-run company does not show the board a wall of leading inputs, and it does not ask a team to steer by a lagging outcome it can only influence indirectly. Match the metric to the altitude of the person reading it.
- Team review: leading inputs, weekly, with owners and targets.
- Board review: lagging outcome, the trend and the narrative behind it.
- The bridge: the tree, which shows how the leaves the team pulls become the root the board reads.
Self-studyDeeper leaves warn earlier2 min read▶
A subtle payoff of decomposition: the further down the tree you push, the earlier the signal. GMV lags. Traffic leads GMV. And "ad impressions" or "email sign-ups" lead traffic. Each level down buys you a little more warning time, because it sits earlier in the chain of cause. That is the real reason to decompose past the obvious - not for tidiness, but because the deep leaves are your earliest alarms. Push the tree until the leaves warn you soon enough to act.
Choosing your leading metrics 7 min live
Not every early number is a good leading metric. The ones worth putting on the wall clear four bars: they are predictive of the outcome, they arrive early enough to act on, they are controllable by a team, and they are owned by a named person. The most common trap is the fourth crossed with the third: a leading metric that looks great on a slide but that nobody can actually influence. That is a weather report, not a steering wheel.
LiveLead time, correlation, and can-we-move-it3 min▶
Three practical questions pick your leading metrics. First, lead time: how far ahead of the outcome does it move? A signal that arrives one day early is barely worth watching. Second, correlation: when it moved in the past, did the outcome reliably follow? A leading metric with no track record is a guess. Third, and most often skipped, can we move it: is there a team with a real lever on it? The trap is a metric that scores well on the first two and fails the third - fascinating to watch, impossible to act on.
- Lead time: enough runway to change the outcome, not just to brace for it.
- Correlation to the outcome: a real, repeated link, not a one-off coincidence.
- Can we move it: a named owner with an actual lever - otherwise it is a barometer, not a control.
The beautiful metric nobody could touch. A leadership team fell in love with a leading index that predicted demand weeks ahead - and spent two quarters reporting it before someone asked the obvious question: what do we do when it drops? Nothing, it turned out. No team owned a lever on it. It was an accurate weather forecast for a business with no roof to fix. They kept it as context and went back to the inputs they could actually pull.
Self-studyThe trap of the uncontrollable leader2 min read▶
It is worth naming why the uncontrollable leading metric is so seductive. It is often the most predictive number you have, which makes it feel like the most important. But importance for a leader is predictive-times-controllable, not predictive alone. A metric you cannot move belongs in the context section of the review, next to the macro climate and the competitor news - useful to know, never something to steer by. Keep your steering metrics to the ones a team can actually turn.
- Predictive but uncontrollable: context. Watch it, plan around it, do not steer by it.
- Controllable but not predictive: vanity, and a gaming risk. Retire it.
- Both: a steering metric. These, and only these, belong on the team's weekly wall.
Five things to raise this week ◐ discussion prompts
- Take the top-line tree from a1 and label every lever leading or lagging. How far up the tree do you have to go before a metric stops warning you and starts just keeping score?
- Look at what your team reviews weekly. Is it leading inputs you can still act on, or a lagging outcome you can only report? If it is the latter, you are managing by rear-view mirror.
- Pick your single most-watched leading metric. Does it pass both tests - predictive of the outcome and controllable by a team? If it fails either, ask why it is still on the wall.
- Find a number your leadership loves that nobody can actually move. Decide honestly whether it belongs in "steering" or "context" - most such numbers are context wearing a steering badge.
- Bring one drop to a3. We will walk the diagnosis together - is it real, which branch moved, which segment inside it, and against what baseline.
Frameworks this session draws on
Metric decomposition has no single certificate - it is a craft distilled from a handful of durable frameworks. This leader track teaches the thinking core of these, applied to the decisions you make. This page draws on:
Three questions before you go 🎯 ◐ 90 seconds
1 · What is the defining difference between a leading and a lagging indicator?
Leading inputs like sign-ups and pipeline move first and give you warning; lagging outcomes like revenue confirm the result weeks later. The gap between them is your window to act.
2 · What is the rule for using the two kinds of metric?
Operating conversations live near the leaves, where you can still change the outcome. The board conversation lives at the root. Manage the leaves, report the root.
3 · A leading metric must pass which two tests to be worth steering by?
Predictive-only is context you cannot act on; controllable-only is a vanity dial that invites gaming. A steering metric has to be both - pull it and the outcome reliably follows.