Choosing the headline metric
State the outcome first, then pick the metric that best reflects that change. A metric that moves more slowly than your planning horizon cannot be the headline one: you will not live long enough to see the effect of your actions.
Three ways to decompose
- Formula. Average deal equals sales divided by orders. Mathematically correct, generates no ideas.
- Additive. Sales equal online plus offline. That is segmentation; it shows no mechanism.
- Multiplicative. Revenue equals enquiries times conversion times deal size. Every factor has an owner and its own set of actions.
What components must be
- they explain the headline metric fully, or nearly so
- logically clean, with no overlap between them
- capable of being broken down further
What drivers must be
- directly controllable: a specific action moves the driver
- locally owned: there is a person answerable for it
- fast feedback: the result is visible in days or weeks
A sales example. Headline metric: revenue. Components: enquiries, conversion, deal size. Drivers under conversion: first response time, share of conversations where price was quoted, follow-ups on stalled threads. All three are controllable and all three are visible within a week.
Six criteria for deciding what to fix first
- the driver's contribution to the headline metric
- how far it has currently dropped
- the upside available
- how controllable it is
- how fast the feedback arrives
- what changing it costs
How to check your tree
- components explain the full 100% of the headline metric
- components do not duplicate each other
- no driver sits at component level, and no segment stands in for a component
- the bottom layer really can be moved by a specific action