Nan Dawkins, Founder/CEO, Serengeti Communications
By Nan Dawkins, Founder/CEO, Serengeti Communications
The Word of Mouth Marketing Association (WOMMA) defines an influencer as “a person who has a greater than average reach or impact through word of mouth in a relevant marketplace.”
Influencers are generally understood to be a crucial factor in creating successful WOM marketing. As a result, much has been written about where to look for influencers and how to engage them. Numerous commercial products offer some form of automated influencer identification, especially for the online environment.
Currently, many of the automated tools that identify and rank influencers rely heavily on the size of the influencer’s following. More sophisticated approaches are emerging, including:
- Analysis of interaction dynamics (i.e., how far an influencer’s post is likely to spread across Twitter or how often, on average, an influencer’s content is shared and propagated by others),
- Analysis of relevance (i.e., is the influencer truly relevant to a particular topic)
- Sentiment analysis (i.e., determining the sentiment of content shared by the influencer)
- Real-time trending of rapidly accelerating influence events (which may point to previously unidentified influencers).
This may sound grandly ambitious and very cool (and it is, in many ways) but the practical reality we face, as a group, is that influencer analysis is very much in its infancy. The companies working in this area (mine included) are facing many challenges, such as the technical issues of dealing with large amounts of data, the difficulties of asking computers to make human judgments about relevance and meaning, and the stark reality of a Black Swan Internet environment which produces influential people and scenarios that could not have been predicted based on past observations (think United Breaks Guitars or any number of other examples).
Simply put: it’s complicated. So perhaps now is a good time to put aside all our network-theory and graph-analysis tools for a moment and consider the possibility that the underlying problem is that we need to view influence in WOM settings as a social fact. Instead, we seem to be framing the concept of influence in a way that assumes it is a natural fact, like pi or Avogadro’s constant. Maybe it’s time for us to go back and read a bit of Émile Durkheim, and then consider suicide.
No, not committing suicide; I mean, suicide as a “social fact.” Durkheim was a French sociologist who made huge news in 1897 with a study that showed suicide rates were much lower in some European religious communities than in others. There was a great rush (at least among academics) to conclude that human behavior might largely be explained in terms of a small number of beliefs and values. More than a century hence, most sociologists now understand that the differences Durkheim found in suicide rates were in fact attributable mainly to the way these different religious communities defined and documented the act of suicide.
Durkheim also pioneered the acceptance of social facts as valid scientific insights, on a par intellectually with the Pythagorean Theorem and E = mc2. But he also pointed out a crucial distinction between social and natural facts: what may be an observably and demonstrably true social fact for a specific social group may not be true for any individual member of that group, and will almost certainly not be true for many in that group. If that seems counterintuitive to you, I agree entirely. But let’s have a bit closer look.
In 2009, the average US family had 3.14 members. I’m going to give the Census Bureau its props and concede that this metric is probably accurate, even though I am confident that I will never, ever find an American family that is exactly that size.
Back to influencers. Consensus on a fully quantifiable, replicable and consistent definition of influence and its impact on WOM may be an ideal worth striving for. But we would do well to consider the possibility that some or possibly all aspects of what we call “influence” may be reflective of influencers in aggregate, but rarely observed on an individual level. So, while it may be true that on an aggregate level, an influencer can be defined as any person with greater than average reach or impact within a relevant marketplace, consider the following:
- Joe has average reach and average influence among a small network. However, Joe’s network has extremely high value to my company. Average influence in an extremely high value network is enough to put Joe on my influencer list.
- John has high reach and influence among a very relevant marketplace of gourmet food lovers. In addition, John has mentioned my brand, (a line of gourmet nuts). Unfortunately, John has a nut allergy. Here is the context in which John mentioned my brand: “Nuts follow me wherever I go. A jar of Brandx gourmet nuts fell on my head while I was shopping in the grocery today and it very nearly killed me.” John might be an influencer in the gourmet foods category, but he probably does not belong on my influencer list.
Another inconvenient characteristic of social facts is that they are fluid. They have a shelf-life and a use-by date. The reality is that the definition of an influencer for a particular brand, product, or company probably will (and should) change based on marketing goals:
- If the goal is to stimulate new, incremental sales (i.e., new buyers), WOM created by less loyal customers who are not opinion leaders, and occurring between acquaintances (not friends) may be more effective (see “Firm-Created-Word-of Mouth Communication: Evidence from a Field Test,” Marketing Science, Vol.28, No. 4 by David Godes and Dina Mayzlin, 2009.
- If the goal is increased brand awareness, someone who mentions my brand often and who has high reach among a relevant audience is an influencer. Someone who has high reach among a relevant audience and mentioned my brand only once, more than a year ago (and rarely mentions competitive brands) may not belong on the target list, at least for this campaign.
These considerations oblige me to consider that, even if a given individual may qualify as an influencer in my product category, he or she might not qualify as a target. And, vice-versa. At the very least, the relative value placed on the criteria used to rank a given influencers’ value (reach, influence, relevance, etc.) may vary significantly based on audience and goals for a particular campaign.
My point? Perhaps we all need to take a breath for a moment and turn our attention to technologies and best practices that discover and incorporate the way different groups define and experience influence. On a practical level, we should at least consider the possibility that the tools are rushing to develop must be flexible — allowing marketers to adjust the dials on the criteria for defining and ranking influencers – in order to be truly useful.
Nan Dawkins is the founder and CEO of Serengeti Communications, a Washington, DC based marketing firm specializing in search, social media, web analytics, digital marketing measurement tools and training. Serengeti recently launched a new social media measurement tool, Social Snap, to provide marketers with in-depth insights into the results of social media programs.
