Important Terminologies: Correlation vs Causation, Input vs Output Metrics, Pareto, False Positives, and A/B Tests

Two metrics moving together need not be cause and effect. Ice cream and sunburn share hot weather; big houses and long lives share the ability to afford healthcare; makeup prices and sales both rising from October to December share festive demand. Discounts normally move inversely to sales, so rising price with rising sales signals demand strength - if you are already ahead of the annual plan, relax discounts and bank margin.

Outputs (revenue, NPS, finish time) cannot be moved directly; inputs can. B2B: meetings, qualified leads, trade shows. Ecommerce: traffic, product-page views, add-to-cart, then conversion. Support: response and resolution speed drive satisfaction. Leading indicators predict future success (on-time delivery -> NPS; daily calories -> weight; new launches -> repeat rate); lagging indicators measure past performance. Optimise leading indicators within a week or two instead of waiting a quarter.

Pareto: 80% of outcomes from 20% of causes - sales from SKUs, revenue from customers, search visits from keywords (prioritise the 500 by volume), industry profit from companies; focus on the 20%, delegate or drop the rest (reused later: 23 of 553 SKUs; 3 of 12 pricing factors). False positive = a positive result that should be negative; false negative the reverse; design for the acceptable error: medicine prefers false positives (retest), courts false negatives (innocent until proven), QC false negatives. A/B test: control sees the default; variants differ; marketing wants the campaign category on top (1.1%), product wants cheapest first (1.15%), control 1% - either beats today, so the brand goal may still pick 1.1%. Report incrementality over control.