Case Study



Rockland Trust’s media mix included multiple one-to-one and offline channels – online video, display, social, search, direct mail, TV, radio, newspaper, OOH. The bank wanted to answer the following questions: (a) How many incremental sales was each channel driving? (b) Was the lift relative to media channel investment efficient? (c) Which combinations of media channels produced a multiplicative effect on sales? (d) What was the optimal investment by channel before sales flattened or declined?


Agent Based Modeling was utilized to quantify lift to investment by media channel, for three separate products – consumer checking, business checking and home equity


  • 18 months of data were carefully entered into the model: bank category competitive and market share; weekly impressions and costs for all media channels; weekly sales by product; brand awareness, consideration and perception
  • Three models, one for each product, were calibrated via a “hold out” – the most recent 6 months of data were withheld, and only the previous 12 months of data were used to forecast sales over the 18 month time period. The resulting trend lines for each product were calibrated to MAPEs of between 5% to 12%


Year over year increase in sales; decrease marketing budget by redeploying to more efficient media channels.

  • Share: