[Feature Request] A more flexible carryover effect

[LukaData]

From our experience, especially, when dealing with larger purchases there tends to be a longer time to convert period, between a marketing spend and the revenue stream. Decisions like buying a car, vacation, appliances, etc. take time. Marketing channels may have the highest impact on the consumer days or weeks after their initial touchpoint.

Similarly many campaigns advertise promotions, discounts, renewals... that will happen in the future. Marketing channel investments for Black Friday start weeks ahead of the actual deals and revenue streams.

Proposed solution: Provide a more flexible approach to the carryover effect such as a delayed adstock function* along with the existing geometric decay.

*Bayesian Methods for Media Mix Modeling with Carryover and Shape Effects 2.1

[alexwakim]

Thank you for your feedback! We'll take your suggestion into consideration.

In the meantime, you might try adjusting the max_lag argument in ModelSpec to change the maximum number of future time periods affected. You could also try using custom alpha_m priors to encourage longer-term effects; an alpha_m prior with more mass closer to 1 will encourage this.