Free Template · KPI Tree

Outcome vs Output
Dashboard.

A KPI-tree template. North star → outcomes → leading metrics → output activities. Stop measuring shipped tickets.

Team: ______
Owner: ______
Reviewed: ______

The trap

Output dashboards (tickets shipped, story points, releases per sprint) measure busyness; outcome dashboards measure whether reality changed. The trap is that outputs are easy to count and outcomes are hard to attribute, so teams default to the easy thing — and then wonder why six months of high velocity didn’t move the business. The KPI tree is the discipline that ties what you can ship to what you’re trying to change.

The 4-tier KPI tree

Every metric belongs to exactly one layer. The layers chain causally — outputs change leading indicators, leading indicators change outcomes, outcomes change the north star.

1North Star
The single numberThe one metric the business is trying to change this year.
2Outcome metrics
3–5 outcomesThe mid-cycle metrics that move the north star. Quarterly cadence.
3Leading indicators
5–10 leadingUser behaviors that predict outcome movement. Weekly cadence.
4Output activities
The workFeatures shipped, experiments run, fixes deployed. Daily/weekly cadence.

Worked example — B2B SaaS Growth team

Growth squad · Q3 · 4 PMs / 8 engineers

1North star
ARR$8.4M → $12M (FY)
2Outcomes
D7 activation rate18% → 30%
Trial → paid conversion9% → 14%
Net new ARR / month$220k → $340k
3Leading indicators
Onboarding completion58% → 78%
Invite-rate per signup0.4 → 1.1
D3 feature adoption22% → 45%
Time-to-first-value11min → 5min
4Output activities
3-step onboarding flow shipped
Invite UX redesign (in build)
Sample-data on signup
First-value moment trigger
Empty-state nudges

Worked example — B2B SaaS Retention team

Retention squad · Q3 · 3 PMs / 6 engineers

1North star
Net Revenue Retention104% → 113% (FY)
2Outcomes
Failed-payment recovery32% → 55%
Seat-utilization rate41% → 60%
Expansion ARR / quarter$140k → $260k
3Leading indicators
Active seats / account3.2 → 4.8
Admin invites sent0.8/wk → 2.4/wk
Pre-renewal product touch61% → 85%
Workspace creation1.1 → 1.9
4Output activities
Smart retry logic (Stripe)
Pre-expiry email cascade
Admin invite flow
Inactive-seat reminders
Renewal-prep digest

The “show me the connection” rule

Every output activity must have a written hypothesis tying it to the layer above it.

Hypothesis formatIf we ship [output], leading indicator [L] moves by [Z%] within [N weeks], which moves outcome [O] by [Y%].

Without a written hypothesis, the activity isn’t measured — it’s just done. Six weeks later, when the outcome hasn’t moved, no one can tell whether the feature was wrong, underbuilt, or correctly built but pointed at the wrong leading indicator. The hypothesis is the audit trail.

Example. “If we ship the 3-step onboarding flow, onboarding-completion moves from 58% to 78% within 4 weeks, which moves D7 activation from 18% to 26%.” That’s a falsifiable claim. If onboarding-completion lifts but D7 activation doesn’t, the leading indicator is wrong — activation is gated by something other than completing the flow.

How to socialize the tree

The tree lives on the wall, in the Slack channel header, in every QBR. The test is whether anyone in the team — engineer, designer, PM, marketer — can recite the layer above their work without looking. If they can’t, the tree isn’t socialized; it’s a Notion page.

Onboarding ritual — every new hire is walked through the tree on day one. They don’t leave the room until they can paraphrase it.
Weekly Slack post — one leading indicator each Monday, with last-week movement.
QBR opening slide — the tree is slide one. Every other slide answers a question the tree raised.
Hiring — PM and EM candidates get asked to describe a tree they’ve worked from. If they can’t, that’s a signal.

The 5-minute weekly ritual

Every Monday, the team lead spends 5 minutes on three questions, then posts the answers in the team channel.

Did the leading indicators move last week? — one number each, vs. last week.
What’s the lag to the outcome? — if leading indicators moved but the outcome hasn’t yet, what’s the typical lag and how much further to wait?
What changes this week? — one decision the tree forces. New experiment, scope change, kill a stalled output.

5 minutes of writing, 30 minutes saved in the standup, 6 hours saved in the QBR.

Reading the tree backwards

The tree isn’t just for planning — it’s a diagnostic tool. When an outcome doesn’t move, walk down the layer below it.

Outcome flat, leading indicator moved. — the leading indicator isn’t actually leading. Replace it.
Outcome flat, leading indicator flat. — the outputs aren’t doing what we said. Either underbuilt, mis-targeted, or scoped wrong.
Outcome flat, leading indicator moved, but only for one segment — the output is right but the addressable population is smaller than assumed. Re-segment.
Outcome moved, leading indicator didn’t. — you got lucky, or the tree is wrong. Audit before claiming credit.

Common KPI-tree failures

Too many metrics per layer — if Layer 3 has 17 leading indicators, none of them are leading. The constraint is the discipline.
No causal hypothesis — outputs listed without the “if X then Y” statement. The tree devolves into a feature list with metrics nearby.
Vanity north star — “monthly active users” with no retention cut, or “signups”, or “page views.” A real north star ties to revenue or retention within 12 months.
Output dashboards rebadged as outcomes — “features shipped per quarter” relabeled “product velocity outcome.” Renaming is not measuring.
Never refreshed — the tree is a living doc. If you haven’t touched it in 90 days, the metrics are out of date and the team has stopped using it.

Refresh cadence

Quarterly — outcomes layer. Are these still the right 3–5 outcomes? Did one stop mattering? Did a new one emerge?
Monthly — leading indicators. Do they still predict the outcomes? If a leading indicator moved 30% but the outcome didn’t budge, replace it.
Weekly — outputs. What shipped, what stalled, what got killed.
Annually — north star. Rare to change, but not never. Strategy shift, segment shift, or business-model change forces a top-of-tree refresh.