Social media has been called “the world’s most effective, yet least understood marketing strategy” (Misner 1999). Marketers have become particularly interested in better quantifying the effects of word-of-mouth communication, as traditional forms of communication appear to be losing effectiveness and new digital channels continue to emerge. Additionally, the inundation of consumer touch-points has created an ability to ‘tune out’ messages, making it even harder to break through the clutter. With the advent of blogs, Facebook, Twitter and other forms of online social communities, understanding and measuring the impact of word-of-mouth is more important than ever.

Academic research, since the 1950s, has demonstrated that consumers respond more favorably to word-of-mouth communication than to a marketer’s paid advertising. However, these studies have historically relied on self-reporting through surveys. With the Internet now the most measurable medium we have ever witnessed, opportunities abound to measure actual word-of-mouth communication on in-market behavior.

However, social media do not work the same way as most paid media channels. Word-of-mouth communication is typically “earned” either by some response to marketing or as a result of some action on the consumer’s part. A consumer may talk about a promotion (s)he has seen in an advertisement, which influences a friend to visit the store and make a purchase. That friend then shares the experience with others, which continues to drive the interdependency between paid, owned, and earned media. As a result, traditional marketing mix modeling methodologies are not able to accurately measure the impact of social media – or any other intermediate or feedback behaviors.

In order to model the direct, indirect, and feedback effects of paid, owned, and earned media on consumer behavior, we have successfully utilized Vector Autoregression (VAR) to accurately reflect how 21st-century marketing and media drive consumer response. Here are a few things we have learned:

• Consumers are indeed more responsive to word-of-mouth than they are to a marketer’s paid advertising.
• However, one way in which word-of-mouth works is to amplify the effects of a marketer’s paid and owned media. If not modeled correctly, total effects of paid and earned media are often underestimated.
• Paid media drives word-of-mouth, which influences consumer response and, in turn, triggers more earned media. Marketing mix measurement methods must be capable of capturing these feedback effects.
• Paid media channels often work differently in terms of their ratio of direct to indirect effects on behavior. Some media work more directly in terms of driving purchase vs. conversation.
• Paid, owned, and earned media differ in terms of the duration of their effects. For example the direct effect of a television ad on sales may endure for a week or two, while the effects of word-of-mouth – triggered by that same ad – can persist for far longer. Most marketing mix models fail to capture these differential time effects.
• Finally, “what” consumers say often matters more than the volume and valence of their conversations. Being able to track the context of the conversation is critical in measuring the impact of social media.
  

In today’s new marketing ecosystem, state-of-the-science analytical solutions are required to accurately measure the return on marketing investment. Vector autoregression has been academically tested and proven to accurately reflect consumer behavior in today’s marketplace.

Please feel free to contact me at craig.stacey@mproductivity.com with questions or comments.

_I have been asked by several people in industry to provide more detail with respect to the points I raised in “A Mix Modeling Manifesto.” While we are working on an academic publication, which could take months (or even years!) to get published, I decided to post some brief notes as follow-ups to my previous article. This is the second installment of those follow-ups.