Following up on my post about the Duncan Watts and Peter Dodds study.

Fast Company has picked up on the discussion, in an interview with Duncan Watts that asks "Is the Tipping Point Toast?". In it, Watts challenges the concept of opinion leaders, particularly the notion that a select number of influential individuals can generate a social epidemic. To do so, he set up a virtual society of 10,000 people who were programmed to be able to influence others, to have varying levels of receptivity to influence, and to have a variety of social connections. Ten percent of the group were made Influentials--meaning they had 40 times the connections of the average person.

He then tried to start a trend in the society and followed how far it went. Thousands of times over.

The experiment generated several hundred cascades (i.e. word of mouth success!). However, the influence of the Influentials was found to be mitigated. A contagion was more likely to be started by the average Joe than an Influential, despite the difference in connectivity. That is, your chances of getting something started are just as likely or more likely if the trend is started by a non-Influential.

That said, the experiment did find that trends that were started by an Influential spread farther than those started by the average person. But these were less likely to get started in the first place.

Watts argues that they best way to start of word of mouth campaign is not to waste time and money targeting a specific group of people. Rather, he advocates "Big Seed" marketing (see his article in the May 2007 Harvard Business Review for his discussion). Essentially, the idea is build a campaign to include word of mouth effects (beyond the ubiquitous forward-to-a-friend) and promote the campaign to as wide of an audience as possible--because you don't know exactly who will start a the trend, the optimal strategy is reach as many as possible.

The ideas presented by Watts have not gone over well with Malcolm Gladwell and Ed Keller, two authors who have benefited greatly from the idea of opinion leaders (see The Tipping Point and The Influentials). Nonetheless, Watts' simulations shouldn't be so readily dismissed as "academic" (as Keller responded).

For one, at least one practitioner who makes living building word of mouth campaigns in the real world has questioned the influence of the Influentials.

Dave Balter, one of the founders of BzzAgent, dedicates a chapter to the "myth of the influentials" in his book: Grapevine: The New Art of Word-of-Mouth Marketing. In it, Balter asserted that everyday, average people can generate word of mouth regardless of their "opinion leader" status. He also noted on at least two campaigns that the profile of individuals who generated the most results were not the ones originally targeted in the campaign.

Another point made by Watts is the idea that you can reverse engineer a trend to understand what started the trend. This completely ignores what happens when something fails, you only see what happens when something succeeds.

Moreover, the success bias ignores the impact of timing. What is successful at one given point in time will not succeed in another because the context and circumstance will be different.

Essentially, his conclusion is that trends are random. Something that those who sell services, knowledge, or ideas based on the ability to create word of mouth don't want to hear.

Labels: duncan watts, fast company, six degrees, word of mouth marketing

In my last post about the Watts and Dodds study, I asked the question (to myself, in all likelihood):

1) If you are building a word of mouth or grassroots communication campaign, do you to take a targeted approach and try to focus on opinion leaders? Or, do you take a scattershot approach and try to reach as many people as possible? (Assume you can't do both, given that your resources are limited).

There are advocates of both approaches. Proctor and Gamble has built in house programs (see Tremor) where opinion leaders are identified, qualified and then engaged through sampling, seeding, referrals, etc. For those without P&G's resources, there are agencies such as BzzAgent that offer a database of opinion leaders for rent, folks who can be targeted with sampling and other programs in an effort to build buzz in support of a product or service.

Others approach it from the direction of customer interaction. Brains on Fire, Church of the Customer, and countless other writers, consultants and gurus assert that the way to generate word of mouth marketing is to be nice to the people that talk about you.

Still others (promotion marketing agencies) maintain that word of mouth uses traditional promotion techniques like seeding, sampling, and so forth.

If you are trying to employ word of mouth for marketing or communication purposes, one of your first steps is to figure out which approach to take. Identifying opinion leaders can be a time consuming and expensive process as you try to build a database of influentials. Going with a service runs the risk of targeting opinion leaders who are not truly leaders in your core market--not to mention the fact that the agency may not even have folks available if your business is in a specialized market. So, how do you optimize you decision?

I think the optimal approach is to create a hybrid approach. Build your own database of opinion leaders--folks from your market and your customer base.

Instead of trying to profile them, let them self select into the database by offering things they want and respond to: information and identity.

Labels: six degrees, social media, word of mouth marketing

A newly published study has raised the question about the importance of opinion leaders in word of mouth marketing and communication. Duncan Watts and Peter Dodds conducted a series of computer simulations that suggest that influentials (a.k.a. opinion leaders) may not be the most important element in determining whether an message spreads throughout a network. Rather, they found that "large cascades of influence" are more likely to be driven by a critical mass of easily-influenced individuals.  Influencees rather than Influentials.

This study raises some interesting questions:

1) If you are building a word of mouth or grassroots campaign, do you to take a targeted approach and try to focus on opinion leaders? Or, do you take a scattershot approach and try to reach as many people as possible? (Assume you can't do both, given that your resources are limited).

2) Isn't this obvious information, that you need a critical mass of easily-influenced individuals for a cascade of influence? Or, does the study offer a valuable new perspective?

3) Can you actually do anything to affect a critical mass of easily-influenced individuals? Or, do you hope that you roam the range looking for a herd?

I hope to tackle each of these questions in forthcoming posts! Feel free to join the conversation.

As someone engaged in the actual practice of interactive marketing, I am interested in what works. There are lots of folks who say they know what works, but I've noticed that few of them are on the client side--working for an organization that is trying reach its market and accountable for bottom line performance.

Those who are selling services (i.e agencies), products (i.e software and web sevice providers) and ideas (i.e. consultants and authors) in the word of mouth marketing space are often hard to pin down when it comes to evaluating the success of their services, products and ideas. Instead of performance metrics, they give you case studies, presentations, and white papers populated with fuzzy math. To be fair, these folks are not on the inside and thus don't have access to the appropriate data. And those who are on the inside often don't have access to the data or can't share it even if they wanted to.

This blog is my attempt to understand more about my field, my place to think out loud. My focus is not about tactics, but more about strategy.

My goal is to post once a week. We'll see how well I hit my numbers.

Labels: interactive marketing, milwaukee