Bayesian regression with adstock. Lightweight MMM's Bayesian algorithm with adstock transformation is the default media mix model option. It accounts for media lags using the adstock method, which is best suited for short-term effects typical of promotional, campaign-driven advertising. It will also generally run the fastest.

Bayesian regression with Hill adstock. This algorithm with Hill adstock transformation accounts for media lags using the Hill adstock method, which is best suited for short-term effects typical of promotional, campaign-driven advertising with non-linear effects. It is more complex but also more flexible than the traditional adstock method. It is especially helpful when you expect to see saturation in your response curves, with diminishing returns on media investment in certain channels.

Bayesian regression with carryover. This algorithm with carryover transformation accounts for media lags using the carryover method. It is best suited to capture the longer lags seen in brand-based advertising and/or complex B2B buyer journeys. It will also generally take longer to run.