The two simplest models are first-touch (the touchpoint that first acquired the visitor gets credit) and last-touch (the touchpoint immediately before conversion gets credit). Both are common defaults and both systematically distort decisions — first-touch overweights top-of-funnel, last-touch overweights bottom-of-funnel.
Multi-touch attribution (linear, time-decay, position-based) distributes credit across multiple touchpoints in a journey. These models are more accurate but harder to implement — they require tracking individual users across sessions, devices, and channels, which has gotten harder since iOS 14.5 and the deprecation of third-party cookies.
Modern best practice is to combine multi-touch attribution at the digital level with periodic incrementality testing (media mix modelling) at the strategic level. Each method's blind spots are filled by the other. Single-model attribution at scale almost always misallocates spend.