Part 5 · 2 chapters · ~12 min

Metrics: North Star, Inputs and Guardrails

Choosing a north star metric, input metrics teams can move, guardrails against damage, leading and lagging indicators, Goodhart's law, funnels and cohorts, and building a metrics tree with engineering instrumentation.

10

A metrics tree

Metrics are a model of the business. Draw the model explicitly so everyone can argue with it.

NORTH STAR, INPUTS AND GUARDRAILS
one metric for value delivered, the levers that move it, and the ones that must not break
north starmonthly active borrowers repaying on timeinputloan approvalsinputrepayment reminders openedinputrepayment success rateguardraildefault rateguardrailcomplaints per 1k usersguardrailcost per loan
swipe the figure sideways, or tap expand for full screen
1/5
north star
One metric that captures value delivered to customers and predicts the business: for a lender, borrowers who repay on time each month (healthy usage, healthy book).
one metric for delivered valuepredicts long-term business health
11

Funnels, cohorts and instrumentation

toolanswers
funnelwhere users drop out of a journey (started KYC → submitted ID → selfie → approved)
cohort retentionwhether users who joined in March still use the product in June, compared with February's cohort
segmentationwhether a change helps some users and hurts others (Android low-end, a country, new versus returning)
event taxonomyconsistent event names and properties (kyc_step_completed {step, method}) agreed with engineering and tested