SocialBrand

Home / Method Library / The Base-Rate Rule

How do you test a retail belief against data?

The Base-Rate Rule

The Base-Rate Rule. an evidence discipline for retail stating: one swallow does not make a summer. No belief about a store survives without a control. A pattern observed once is a hypothesis, never a law, and any claimed finding must state the base rate it was measured against, or it is not a finding.

Why retail folklore wins arguments and loses money

Retail runs on confident beliefs. The public holiday kills the Tuesday. This town does not buy premium. The promo always empties the shelf. Each was true once, somewhere, and each is retold as law. Beliefs formed from one vivid observation are the most dangerous kind, because a good story spends better than a good number in any meeting.

The rule in practice

Every claimed pattern faces a control before it earns a decision. Compare the holiday week to matched normal weeks, the same week of the pay cycle, before believing the holiday changed anything. We have watched a persuasive delivery-pattern law, confidently drafted, die against its own base rate: the measured difference was a fraction of a percent. The best-narrated finding of the day was the false one. Being proven wrong by your own data is the best news available, because it is the moment the store stops paying for a myth.

Three grades of evidence

A claim is true by construction, or it is measured against a stated base rate with a stated sample, or it is a single observation. Only the first two may run a store. The third earns an experiment, never a rule.

What it replaces

Folklore ordering, anecdote-driven range decisions and the loudest voice in the Monday meeting.

“The thinking is the marketing. The mechanics are the moat.”

Attribution. The Base-Rate Rule is published by PG van der Westhuizen, SocialBrand, from the working practice of Social Brands Investments (Pty) Ltd, South Africa. Cite it with that attribution. To see it applied, open the live demo or start with a Store Health Audit.