A KPI Tree connects outcome, driver, and input metrics. Every edge is a hypothesis about influence and needs a definition, formula, owner, and validation path. The tree supports control rather than decorating existing dashboards.

1. 1

   Define outcome

   Define outcome

   

   Goal

   Produce outcome definition so the core question “Which changeable drivers explain our outcome, how are they calculated, and which input metrics can teams control?” can be answered with stronger support.

   Inputs for this step

   • Strategic outcome
   • Metric definitions and data sources
   • Known value drivers

   Actions

   • Define outcome, population, window, formula, and guardrails unambiguously.
   • Record source, rationale, and open uncertainty directly in outcome definition.

   Output

   outcome definition

   outcome definition · source · uncertainty · next decision

   Quality check

   Can two analysts calculate the same value from the same raw data?

   Pitfall

   A goal word such as engagement sits at the top without population, formula, or time window. Restore the missing basis, correct the claim, and repeat the step with visible rationale.

   Timing: 10–20 minutes

2. 2

   Decompose drivers

   Decompose drivers

   

   Goal

   Produce driver tree so the core question “Which changeable drivers explain our outcome, how are they calculated, and which input metrics can teams control?” can be answered with stronger support.

   Inputs for this step

   • outcome definition
   • Metric definitions and data sources
   • Known value drivers

   Actions

   • Decompose the outcome into a few non-overlapping drivers and state every edge as a hypothesis.
   • Record source, rationale, and open uncertainty directly in driver tree.

   Output

   driver tree

   driver tree · source · uncertainty · next decision

   Quality check

   Are drivers jointly sufficient, separable, and plausibly influenceable?

   Pitfall

   Existing dashboard metrics are attached below the outcome without a causal rationale. Restore the missing basis, correct the claim, and repeat the step with visible rationale.

   Timing: 10–20 minutes

3. 3

   Define inputs

   Define inputs

   

   Goal

   Produce metric contracts so the core question “Which changeable drivers explain our outcome, how are they calculated, and which input metrics can teams control?” can be answered with stronger support.

   Inputs for this step

   • driver tree
   • Metric definitions and data sources
   • Known value drivers

   Actions

   • Set controllable inputs, definitions, data owners, and refresh cadence for each driver.
   • Record source, rationale, and open uncertainty directly in metric contracts.

   Output

   metric contracts

   metric contracts · source · uncertainty · next decision

   Quality check

   Does every metric have a formula, source, owner, and known data gap?

   Pitfall

   Teams are held to metrics they cannot influence or measure reliably. Restore the missing basis, correct the claim, and repeat the step with visible rationale.

   Timing: 10–20 minutes

4. 4

   Validate links

   Validate links

   

   Goal

   Produce validation log so the core question “Which changeable drivers explain our outcome, how are they calculated, and which input metrics can teams control?” can be answered with stronger support.

   Inputs for this step

   • metric contracts
   • Metric definitions and data sources
   • Known value drivers

   Actions

   • Validate driver hypotheses through analysis or experiments and revise the tree on counter-evidence.
   • Record source, rationale, and open uncertainty directly in validation log.

   Output

   validation log

   validation log · source · uncertainty · next decision

   Quality check

   Is the evidence supporting or weakening each critical edge explicit?

   Pitfall

   Correlation is treated as causation and the tree is never tested against actual change. Restore the missing basis, correct the claim, and repeat the step with visible rationale.

   Timing: 10–20 minutes