Maturity is neither a trophy nor an average of good practices.
Maturity is neither a trophy nor an average of good practices. It identifies which decisions an organisation can take repeatedly, with which evidence and controls. A team can have strong governance and skilled analysts while unable to optimise automatically if data do not reconcile or recommendations are never used. Operational level is constrained by the weakest critical capability for the intended decision.
Decision and method.
Authorise diagnosis, experimentation, allocation or automation separately from minimum thresholds by capability, not a compensatory global score. List real uses—describe, compare, experiment, forecast, allocate, automate—and state amount at risk, frequency and reversibility. Assess five capabilities on recent observable evidence: data dictionary, decision register, experimental protocol, contribution account, independent validation or proof of use. A claim without document, log or reproducible result is intention, not capability. Define use-specific gates: authorised level is the one where every critical gate passes; remaining capabilities drive the progress plan. Address the capability limiting the highest-value decision with owner, deliverable, date and acceptance test.
Worked example.
Organisation scores governance 4, data 2, experimentation 3, economics 4, adoption 2; average 3/5. Yet only 72% of exposures reconcile to sales and 40% of documented recommendations are used. Descriptive diagnosis requires at least 2 everywhere, so it is allowed. Allocation requires data 3, economics 3, adoption 3: two gates fail. Automated optimisation requires 4 in data, governance and control, so it is blocked despite average. Priorities are reconciliation keys and a decision register; models compare scenarios and design tests, not automatically move budget.
Checks and limits.
Levels rest on recent, verifiable evidence; each decision type has its own minimum thresholds; weak critical capability cannot be averaged away; the progress plan targets the next useful decision with acceptance test. A generic grid must fit organisation risk and size. Process presence does not guarantee execution quality, and maturity score does not directly measure marketing performance.
Resources and sources.
Download the maturity grid. Related: test data quality, govern models, contract KPIs. Sources: ISO 9004:2018; Davenport and Harris (2007); Keeney (1992).