The design half starts with a customer profile built from fieldwork, not a conference room. Jobs: what the customer is trying to get done, whether functional (transfer money), social (look competent), or emotional (stop worrying). Pains: what blocks or annoys them on the way, ranked by intensity. Gains: outcomes they'd value, ranked by relevance. The ranking is the discipline. A canvas where every pain matters equally is a canvas nobody interviewed for.
The value map answers the profile: which pains do our features actually relieve, which gains do they create, and which do we refuse to address? Fit means your relievers hit the pains customers rank highest, not the pains you happen to be best at relieving. This is the Beginner lesson about sacrifice, now with a measurement attached: a proposition addressing the third-ranked pain brilliantly still loses to a rival mediocrely relieving the first.
The testing half is an evidence ladder. Interviews surface language and rankings but can't predict behaviour (the say–do gap). Smoke tests (a landing page, an explainer, a 'fake door' button for a product that doesn't exist yet) buy behavioural signal for days of effort. Concierge and Wizard-of-Oz pilots deliver the promise manually to a handful of customers before any build. Pre-orders and deposits are the truth serum: money is the only survey answer that can't be polite. The rule: test the riskiest assumption first, not the easiest.
Knowing when you've passed matters as much as testing. Behavioural thresholds beat applause: retention curves that flatten instead of decaying to zero; repeat purchase without prompting; the Sean Ellis question ('how disappointed if this disappeared?') clearing roughly 40% 'very disappointed' among real users; willingness-to-pay confirmed by paid conversion, not stated intent. A proposition is proven when behaviour repeats, never when the focus group smiles.