As the one-time CMO of a large global wealth management business with severe reputational issues during the 2007 financial
crisis, I struggled mightily with how to engage with and use social media to help my brand.
Five years later, you’d think things would be better. But almost all of the CMOs and Marketing leaders I talk with are still struggling with social media. Everyone knows it’s important, they know they need to engage with it, yet they are still trying to answer fundamental questions that are relevant to any Marketing activity:
- How do I engage and use social media to benefit my brand and drive sales?
- How do I measure social media and the impact it’s having on my brand – positive or otherwise?
These are challenging questions that almost all Marketers, no matter their sophistication, are struggling with. The question is, why?
Social Media as a Business Driver
In most Marketing activities, Marketers design marketing programs they believe will increase basic Marketing measures like awareness, trial, and consideration and ultimately result in higher sales.
To do so, Marketers have created predictive tools that increase their confidence that these programs will ultimately work. A few examples include:
- Copy testing to predict likely advertising success in building sales.
- Purchase intent scores, which measure the likelihood of consumers buying your product.
- Market mix modeling norms which indicate how much sales $1 of investment will yield.
A key issue with social marketing programs is that there are few of these measures and even where they exist, they are highly suspect. Here’s why.
The Impact of Social – An Experiment
In his book ”Everything is Obvious Once You Know the Answer,” Duncan J. Watts describes a web-based experiment designed to emulate a ”market” for music as a means of measuring the impact of social media. He designed the experiment for consumers to listen to, rate, and potentially download songs.
First, he created two groups. Both groups were given the same set of songs to listen to and rate. Each group was further divided into subgroups so it would be possible to see how multiple groups responded to the same environmental conditions.
But there was one key difference between the two groups. Group A consumers were not able to see how many of their compatriots were downloading the songs they were listening to. Group B, in contrast, was able to see the downloads and rank order popularity of the music based on other participants behavior. Said differently, group B had a ”social layer” of potential influence while Group A did not. The experiment was ”repeated” multiple times given that each of the subgroups had different participants.
Group A results with different participant groups were remarkably consistent – the rank order of songs was similar across groups. If song A was number one in the first group, it was also number 1 in the 2nd group, 3rd group, etc.
This tells us not only that consumer music popularity without a ”social layer” is consistent and relatively unchanged across groups, but it is highly predictable.
Group B results were quite different. The most popular songs, and the less popular ones, were typically even more popular or less popular in Group B than Group A – e.g. the ”social layer” of seeing how other participants downloaded songs sharply accentuated the way the degree to which consumers in Group B rated and downloaded songs.
But equally important, results across different participant groups showed large variations in popularity. So, if song A was most popular in the first group, it might only have middle popularity in the 2nd group and so forth.
Social Makes Results More Extreme and Unpredictable
Duncan Watts had this to say about the results:
”In all the