Brad FayEd KellerRick LarkinKoen PauwelsDeriving ValueFromConversationsAbout Your BrandResearch shows that both online and off-line customerconversations drive purchase decisions — but they requireseparate marketing strategies.Reprint #60210https://mitsmr.com/2zEc7M2Most studies on social media marketing effectivenesshave looked at how brand engagement on specificplatforms such as Facebook or Twitter (for example, thelikes, shares, retweets, and comments) responds tomarketing initiatives as opposed to considering the socialecosystem as a whole. There is little research looking atoff-line conversations — those that occur face-to-face atthe office watercooler, over the kitchen table, or at ahealth club — because of the difficulty and cost ofmeasuring them. However, we addressed that challengeby asking selected consumers to recall the product andservice categories and brands they talked about the daybefore, including whether the brand conversations werepositive or negative. We examined this survey data foroff-line conversations along with social media data sothat we could compare the two types of conversations andidentify trends in both. We also looked at weekly adexpenditures for specific brands from Numerator, anadvertising tracking company, and sales data to create acomprehensive picture of the factors that lead consumersto buy certain brands. (See “About the Research.”)AAbboouut tt thhe Re ReseseeaarrcchhFor our research, we developed a proprietary dataplatform to incorporate online and off-line conversationdata on 501 U.S. brands. For the analysis presented here,we collected online data for 2015 and 2016 and off-linedata for 2013 through 2016. Online data was collectedthrough key-word searches of Twitter, public Facebookposts, blogs, forums, and consumer review sites. Usingnatural-language processing, we analyzed whether theconversations were positive or negative. Our continuoussurvey research program yielded data on brands from anaverage of 7,000 off-line conversations per week withconsumers ages 13 to 69. Respondents were asked toreport on whether their conversation about each brandwas positive, negative, neutral, or “mixed.” Our initial stepwas to correlate the online and off-line data streams forall brands. We then did a regression analysis to link theonline and off-line conversations to third-party weeklysales data that we acquired for 175 brands, and to weeklyad expenditure data for a subset of 21 of those brandsusing a method known as market mix modeling. iWe used this approach to study the relationship betweenonline and off-line conversations in 15 industries,including electronics, packaged foods and beverages,telecommunications, finance, and travel. For many of the500 brands, we were able to obtain third-party sales data,and we paid particular attention to a subset of 21 brands— including Apple, Intel, A&W, Budweiser, Campbell’s,Lay’s, Pepsi, Red Bull, and Revlo...
