Writing in the current issue of Admap, Matthew Mee (global chief strategy officer, MediaCom), Nigel Shardlow (director of planning, Sandtable) and James Allen (senior data scientist, Sandtable), outline an approach that combines respondent-level category purchase data with media spend data to create an individual-level simulation of market dynamics.
“The model provides us with an empirically grounded way to answer the following question, for a given category, and a given brand: given the characteristics of the targeted advertising I’m able to buy, how much should I add to the mix in order to get the best bang for my buck?”
Using this method, they compared the effect of targeting in two different CPG categories – laundry and men’s razors – and reported that targeting makes more sense for razors than it does for laundry.
“Including media that targets system buyers (i.e. people who buy razors with separate handles) alongside broadcast yields additional sales for [razor] Brand G, but tops out with targeted media making up about 3.5% of the media mix,” they write.
“By contrast, including media that targets laundry buyers alongside broadcast normally depresses sales of [laundry] Brand P from the start, and even for the best target group can only produce a marginal uplift.”
And, they add, “the downside of overspending on targeted media is much worse for razors than for laundry, due to saturation effects in the smaller target groups”.
When considering responsiveness, the authors observe that this is highly dependent on category.
A good targeting algorithm, they suggest, can be expected to reach a more responsive population first, at which point it will outperform broadcast media; but as the accuracy of the targeting decreases, its response will fall closer to the broadcast line.
“The big question of how much targeted advertising is too much is an empirical question,” they conclude – and one that can only be answered on a category-by-category basis by bringing data together with a plausible individual-level model of how categories work and respond to advertising.
Sourced from Admap