“It doesn’t get more timely than this paper. Online advertising continues to grow at a very fast pace, in large part because of better targetability and measurability. As far as research methods go, it has given A-B testing a new lease on life, as marketers feel they can test everything before spending on it. But every test needs a control group and control groups cost money and/or they are not really equivalent to the treatment groups, given the algorithms used to deliver ads.

“The authors present the concept of Ghost Ads – a way for marketers to get a more valid and more precise estimate of the actual effect of their ad at a lower cost. Often, control groups are shown a public service announcement (PSA). Instead, as the authors write, ‘ghost ads identify ads in the control group that would have been the focal advertiser’s ads had the consumer been in the treatment group; as such, ghost ads deliver valid estimates of the average treatment effect on the treated…. Instead of PSA ads, the control group consumer sees whatever ads the ad platform chooses to deliver when the focal ad is absent.’ If you spend substantial sums on online advertising, check this paper out. I found the idea very interesting, and, if it can be implemented, the benefits seem to be substantial.”

“This is a quick read with a simple but important take-away. Firms want consumers to write reviews. One way they might encourage this is by offering monetary rewards. Apart from any credibility and undesirable signaling problems that doing so might create, this paper shows that monetary rewards don’t even achieve their main purpose. In a field experiment, the authors find that a monetary reward increases contributions manifold (1400%) from members of an online community that are not socially-connected (i.e., have few ‘friends’ in the community) but decreases contributions by 90% from more socially-connected members. Because more connected members have a much higher baseline probability of posting reviews, the net result of offering the reward was a significant decrease in contributions!

“Lots more to study, of course, such as what happens with higher reward amounts, what impact, if any, there is on the content of reviews, etc. But, this study is enough to give serious pause to the idea that you can just pay consumers to generate content – there is important heterogeneity in response here as in all other marketing activities.”

“This paper is the segue between my timely and the timeless selections. We’ve been studying the importance of Customer Lifetime Value and the effectiveness of targeted campaigns and CRM for many years now. This paper reminds us to connect (pun intended) social connectedness and CRM by showing that the effect of a campaign extends to the non-targeted consumers within the target consumers’ social network. Basically, these non-targeted consumers consume more and churn less because of the increased communications between them and the targeted consumers who they have tight social connections with. In the field experiment that the authors analyze, what they call the ‘social multiplier’ is 1.28 on average. In other words, the effect on non-targeted connections is 28% of the effect on targeted consumers themselves. The lesson of the story – target consumers with promotional and other CRM campaigns not just based on their profitability but also based on the strength of their networks.”

“The saying that ‘I know only half of my advertising works but I don’t know which half’ has become trite with over-use. We have accumulated several empirical studies of advertising effects but unfortunately, we haven’t gotten much closer to finding out which half (if that!) of advertising works, at least for mass media advertising. This meta-analysis doesn’t get us all the way there either but it does provide several very useful insights.

“What is new compared to previous meta-analyses of advertising? It compares mass advertising with targeted advertising and with personal selling, it correctly gets at both the share of the total effect that is longer-term, it provides an updated estimate of the duration of the long-term effect, and it highlights important budgeting implications for marketers and modeling implications for researchers. Read the paper for details, or at least study Tables 5, 6, and 8 carefully.

I found it particularly important to compare advertising (which is often over-emphasized by marketers) and personal selling (which is seriously under-studied) and see how much longer the effects of the latter last; I also found it very useful to have a validly arrived at estimate of the duration for which advertising’s effects typically last. And I was surprised at how much influence the functional form of the model has on the estimated long-term effects.”

“Distribution is the most important element of the marketing mix. We know its effect on sales and share far exceeds that of, say, advertising. But, it isn’t clear yet whether there are increasing or decreasing returns to distribution. Paul Farris and colleagues revealed a well-known empirical generalization– the cross-sectional relationship between distribution (%ACV or %PCV) and market share is convex for consumer packaged goods, both at the brand and individual SKU level. They also offered several possible reasons for the convexity but cautioned against inferring causality from the association.

“This paper shows that the convex relationship is causal, at least in the empirical context studied by the authors. In other words, the percentage increase in sales for an SKU for a given increase in distribution is bigger at higher levels of distribution. The authors demonstrate that this convexity is largely attributable to one of the mechanisms identified by Farris and colleagues – SKUs first get distribution in large stores with large assortments, but the last few points of distribution are in stores with smaller assortments. In these smaller stores, the SKU gets a much larger share of the pie even if the total pie is smaller. I like this paper because of the level of detail at which it can track distribution changes – in different kinds of stores. We would not be able to discern such increasing returns to distribution if we were only studying the large assortment supermarket and mass formats for which scanner data are usually available.”