Glossary of marketing-mix decisions
Terms are defined through their role in a decision, their unit of measurement and the interpretation error they help avoid.
Adstock
persistence of a media effect after exposure; its half-life must fit the channel mechanism and data.
Attribution
a rule allocating an observed outcome across contacts or channels; it assigns conventional credit and does not prove causality.
Cannibalisation
value from a new offer that displaces sales an existing offer would otherwise have made; assess displaced contribution, not volume alone.
Causality
an intervention changes an outcome relative to a defined counterfactual. The counterfactual is never observed directly and needs randomisation, a comparable group or an explicit model.
Counterfactual
an intervention changes an outcome relative to a defined counterfactual. The counterfactual is never observed directly and needs randomisation, a comparable group or an explicit model.
Response curve
estimated relationship between investment and incremental outcome, including thresholds, diminishing returns and an observed range. Do not extrapolate saturation far outside that range without a stress test.
Saturation
estimated relationship between investment and incremental outcome, including thresholds, diminishing returns and an observed range. Do not extrapolate saturation far outside that range without a stress test.
Price elasticity
relative change in demand associated with a relative price change, for a defined unit, period and context.
Uncertainty interval
values compatible with data and model assumptions; it does not guarantee that measurement or selection bias has disappeared.
Incrementality
incrementality is the difference caused by an action against its counterfactual. iROAS relates causal economic outcome to causal spend; both depend on the counterfactual, horizon and currency unit.
iROAS
incrementality is the difference caused by an action against its counterfactual. iROAS relates causal economic outcome to causal spend; both depend on the counterfactual, horizon and currency unit.
Marketing mix
the marketing mix coordinates controllable offer, price, distribution, communication and resource decisions. Marketing Mix Modeling is an aggregate time-series model with transformations, controls, uncertainty and validation.
Marketing Mix Modeling
the marketing mix coordinates controllable offer, price, distribution, communication and resource decisions. Marketing Mix Modeling is an aggregate time-series model with transformations, controls, uncertainty and validation.
Bayesian prior
a prior expresses information available before model data; marginal return is the expected result of the next unit invested; triangulation compares methods with different biases to identify robust conclusions and needed tests.
Marginal return
a prior expresses information available before model data; marginal return is the expected result of the next unit invested; triangulation compares methods with different biases to identify robust conclusions and needed tests.
Triangulation
a prior expresses information available before model data; marginal return is the expected result of the next unit invested; triangulation compares methods with different biases to identify robust conclusions and needed tests.
Customer lifetime value
expected future customer margin, net of service costs and weighted by retention; it depends on the horizon and is not cumulative revenue.