Execute growth interventions

A Play is a repeatable commercial intervention with defined triggers, evidence, inputs, workflow, decision rights, measurement and learning. A campaign idea, prompt or technology feature is not a Play. Every Play includes a basic version, an advanced option, a scale rule and a stop rule.

Acquire customers whose incremental contribution can justify acquisition and servicing cost. Trace relevant intent through proposition, verified conversion, activation and cohort quality. Preserve native platform delivery where it supports the objective; improve the quality of the information supplied to it. An advertiser can own economic definitions without building an auction engine. [STR §§12,16; GRW §§8–24]

P03 starts with source-to-customer-to-contribution linkage and a narrow intent or value-signal intervention. P04 tests whether material attributed acquisition is incremental. P05 tests a meaningful proposition, concept or proof point. Use P02 first when the true problem is a broken form, unclear price, unavailable product or delivery constraint.

Demand creation and demand capture have different lags. Branded search and direct visits can be useful indicators of future demand but are not themselves causal proof. Do not require every brand investment to pass a short-window click ROAS hurdle. AI-generated content volume cannot replace distinctive customer value and substantiated claims.

Define the first event in which a customer experiences product value, separately from a signup or administrative completion. Find where people fail to reach it. Repair product and service friction before adding reminders. A useful onboarding message is one possible treatment, not the default diagnosis. P06 connects Product, Lifecycle and Service under one accountable outcome owner. [STR §13; GRW §28]

The activation milestone may predict retention without causing it. Track later cohort contribution and complaints so a cosmetic activation gain does not masquerade as growth.

Ask whether the cause of churn is addressable and whether an available treatment changes the outcome. A high-risk customer may be impossible to retain profitably; a low-risk customer may accept a discount while behaving exactly as before. Compare service recovery, product repair, education, an authorised offer and no treatment. P07 makes total assigned-population contribution the economic outcome. [STR §14; GRW §§18–21,29–31]

Apply mandatory eligibility and contact suppression consistently. Test economic suppression only where legitimate. Never randomise customers into a prohibited contact condition to estimate its commercial value. Keep discount leakage and service workload inside the value calculation.

P08 covers usage, cross-sell, upsell, win-back and referral as variants of one question: which relevant intervention adds net account value? Require a genuine customer need and suitable product. Include cannibalisation, switching, repeat churn, capacity, fulfilment, rewards and fraud where relevant. Referral codes record attribution; they do not prove the referred customer would otherwise be absent. [STR §15; GRW §§32–36]

In complex B2B, use account-level evidence, sales and customer-success judgement, longer outcome horizons and careful treatment assignment. AI may prepare verified account research; volume-led autonomous outbound is not the operating model. In transactional businesses, separate frequency, units, price and unit contribution. In regulated contexts, keep qualified human review for consequential offers and claims.

P09 turns a bounded customer or market question into cited evidence and an actionable brief. P10 redesigns one repeated workflow and tests whether AI adds value after process repair. Either can be useful without causing a new customer intervention; then judge it on accepted-work quality, time and real cost, and do not invent a revenue benefit.