- A company which for years had relied on marketing mix models to attribute sales volume to marketing plans was challenged with the arrival of new marketing channels and the ways to measure them. They wanted to expand their traditional approach, which measured things such as consumer and trade promotions and television advertising, to include social media, internet banner ads, digital coupons, and online search.
- We started with a sophisticated, yet traditional econometric market response model that included Bayesian statistical principals and adapted it for the kinds of sales responses suitable for these new media. The input data itself for these sources required different thinking and different transformations than traditional measures of marketing influence.
Among the findings:
- Volume attributed to social media can be as high as 5% but usually much lower. And most of it is “unearned” in the sense that the company does not control it.
- Internet banner ads can be around 2.3% of volume and digital coupons typically 0.5% or less.
- There is a fairly large interaction between television advertising and social media, and the interaction between TV and search is about 75% of that of the TV-social media interaction. But TV does not drive very much digital coupon use, and no increased click-through of banner ads was detected.
- Connecting the data from these new sources posed more of a challenge to the company than expected, and decisions needed to be made in how to invest in the new data sources and the infrastructure to incorporate it.