Debating the Influencer model: Fast Company debates the "Un-Tipping Point"
Several people sent me this article recently knowing my interest on the topic of influentials and their role in networks.
[Ironically, reading some of the discussion this article has spurred makes me think that the article that de-bunks the influencer model actually re-enforces the influencer effect on the topic of how information flows.]
I haven’t made it a secret that I have my own effort underway to author a book on the topic. In my role the past 5 years at Microsoft I’ve been a central figure in developing, leading and advocating for the importance of finding, thanking and engaging influentials/mavens/experts/advocates/enthusiasts. I’m guilty of co-mingling these terms which is, I think, contributing to the hubbub about this topic. The practical side of me resists the semantic debate of tearing these terms apart and trying to define the parameters of each. For one thing, it’s been done before and didn’t seem to resolve the debate. For another, it adds complexity which makes understanding the principles and the actions to be taken more difficult. Now, I’m not advocating laziness here! This is a complex topic - the issue is in “making it simple, but not simplistic.”
The article focuses on the seemingly opposite views of experts on either side of the influencer debate…and in all fairness, an article that supported the influencer model would hardly be worth publishing - as the prevailing view, writing in favor would hardly be newsworthy. The primary focus of the article revolves around a profile of Duncan Watts (recent Yahoo researcher and network-theory scientist on sabbatical from Columbia University). The net of it (which is overly simplified) is that networks matter, but influencers within the network have at best a random impact on adoption/trend launching. And giving the choice between trying to target market to influentials vs mass marketing that mass marketing is the right choice. Duncan provides a fair amount of research and simulations to back up his views and I’d say his research is worth reading and understanding. Not to mention he is colorful in his disagreement with Gladwell. Find more on Duncan here.
On the “other side” of the debate, the columnist interviewed Ed Keller of Keller Fay and co-author of The Influentials. This portion of the article highlights a “heated rebuttal” between Keller and Watt’s that occurred last fall at an ARF event.
Before I go on…some disclosure: I was at that ARF event as a panelist who followed Ed and Duncan’s “heated debate.” I’m not sure how or who characterized this as a heated debate. I think that was part of the intent of the organizers - but it was hardly heated. It wasn’t heated by west coast standards, much less for New York! Further disclosure, I have met and co-presented with Ed a few times and have met and talked with Duncan. I’m also a co-chair of an influencer council recently started by WOMMA - so I have intersected with these individuals and the topic many times. I was also interviewed for this article by Clive Thompson, but didn’t make the story - I understand why - I wasn’t talking about marketing.
My panelist response to the debate was essentially that I disagree with both of them (which also means I agree with both on some points). For those that have read my blog for awhile, none of this will come as a surprise.
Here’s the deal. Defining influencer strategy with the narrow end-benefit of driving trends/adoption with push based marketing I think is the wrong thing to do. Whether that alone will work or not probably depends more on how good the product is than on who or how you target. This is not what interests me around the topic of influentials and I think constrains the research parameters in ways that generate multiple truths. For example - in Duncan’s simulations, the assumptions are around what I’ll call untouched consumers (they had no previous interaction with the product). In this scenario, I think Duncan is right, it is luck if you get something to go viral. Influencers (even in Duncan’s research) can extend the reach, but can’t guarantee success - there are too many other variables in the system so all you can prove is that success is random. (Note: Nothing guarantees success, it’s about increasing probability).
In Ed’s extensive research, he’s demonstrated (and documented in a recommended read: The Influentials) that an elite 10% of participants in communities are 2-5 times more likely to engage in advice-giving conversations. Thus, given limited investment dollars to drive word of mouth, influencer based outreach should be a fundamental part of any thoughtful campaign. Ed also rightfully reminds us that in business there are no guarantees and that strategies that increase probability are generally good business choices vs research simulations looking to pave the final mile.
The research I’ve been a part of certainly supports a core part of Ed’s conclusion about the most engaged 10%. We’ve seen time and again what I’d simplistically describe in the following taxonomy for participation in technical communities (1/9/90):
- <1% of unique participants in a community are essentially “answer people” and contribute in extremely disproportionate quantity (both pure volume and average days active per month). The side bar here would be that quantity does not always = quality, but again, statistically those that are not contributing quality don’t sustain at this level over time - it is largely self governing and not difficult to parse the noise from the value based contributors.
- ~9% of unique participants demonstrate similar behaviors as the above elite answerers, but at more modest levels. This correlates reasonably with Ed’s 10%.
- ~90 of unique participants lurk/contribute.
- note: This taxonomy is admittedly a bit simplified. I blogged a deeper opinion on taxonomy some months back here.
While the majority of my experience is with tech communities, fellow community managers across many different areas have reported the same general distribution.
This is where I’ll circle back to the trouble with semantics. What does an influencer influence? Product awareness/sales? Product usage? Product innovation? An influencer strategy that is designed for maximum benefit has to encompass all of these functions. As a I wrote earlier, I’m not a fan of the term “influencer marketing” as it generally is used to describe a catalyst for Word of Mouth that assumes information/communication flows in one direction from A to B to C, etc. This may or may not be the case, but if you want to increase probability (as well as many other business benefits) the flow has to be bi-directional.
Specific example: What if your engaged influencers never told anyone about your products, but gave you 10X the feedback on how to improve the quality or relevance of your products. Is the influencer model wrong? No - the business goal was different - or at least more complex than one dimension focus on buzz generation.
Net net: Duncan’s research is interesting and I plan to continue to follow it, but it feels apples to oranges to compare that research to the world of influencer engagement that I support.
Sorry for the long post, I doubt I have yet simplified this topic as of yet…more to consider.
Popularity: 80% [?]