Ok, I've been away for a while but I thought a good post to resume with would be to review a recent article on the theory of "big seed marketing" written by Professor Duncan Watts, Jonah Peretti, and Michael Frumin. A version of this appeared in the Harvard Business Review, but my critique will focus on the version available at the Columbia University web site (which provides more detailed results and is referenced in the HBR article).
First, let me provide a quick summary of the article's main points, and then I'll discuss what the "news" is here, why this article could be helpful to organizations wanting to better understand marketing communication, and three limitations of the article.
Let's start with the summary:
The authors start by pointing out that viral marketing is seductive and often presented as a panacea: you just need to pick a small number of people to seed an idea and watch it go "viral"; oh, and by the way, all of this can be done on a small budget.
But, the authors caution, for as many viral marketing successes out there, there are just as many, if not more, failures. It's just plain hard to consistently create media that will "take off".
The alternative, the authors contend, is Big Seed Marketing (which I'll abbreviate as BSM). BSM "combines viral marketing tools with old-fashioned mass media in a new and creative way." Its purported benefits are risk-reduction and greater predictability than just viral approaches.
To explain BSM they start with a distinction between mass marketing and viral marketing. Mass marketing can be summarized by the formula n = pN, where n is the number of conversions (positive behaviors or attitude formation, such as purchasing product or forming a favorable impression), p is the probability that exposure to a the ad will result in a conversion, and N = the number of people exposed to the ad. To get more conversions, a company either increases the number of people to which they expose the message, or increase the probability of conversion (for example, by making a better -- more informative, interesting, memorable, etc. -- message).
In contrast, viral marketing starts with a small number of seeds who will then share the message with their friends, who will then share with their friends, etc. (I'll discuss below how this definition actually conflates "influencer marketing" and "viral marketing", but we'll work with this definition for now). Expressed as a formula, viral marketing is R=Bz. R is the "reproduction rate," or the rate at which new converts are generated from existing ones. B is the probability that a WOM transmission occurs (which can be based on a number of factors, including the number of times a person needs to hear a message before passing it on, memory, etc.) and z is the number of people told.
The whole theory of viral marketing is based on epidemiology, or the study of how diseases spread (or don't spread) throughout a population. If the reproduction rate of a disease is greater than one (R > 1), more and more people are infected at each generation, or degree, removed from the initial person infected, and the disease spreads. If less than one, the disease eventually peters out over time because fewer than one person at each degree is infected. Applied to marketing, a reproduction rate greater that one means that more than one other person, on average, is told (R > 1). If that reproduction rate continues at each subsequent generation, then an idea will go "viral" because one person tells more than one person who tells more than one person, etc.. If R < 1, then the message will decay, over time. R = 1 is the threshold; anything above 1 and the message will spread over time, while anything below 1 and the message will decay over time.
But the authors point out a "fundamental flaw" when applying the viral analogy to marketing -- specifically, diseases, unlike many large organizations, don't have have the resources to use the mass media as a way of seeding the message to large numbers of people.
So, the idea of BSM is to start the message with a large number of people and provide tools for people to share the message with others. Even if the reproduction rate is less than 1 (again, meaning that the person who experiences the message passes it on to less than one person on average), a large number of people will still be reached as the message will take a while to decay).
The authors present examples from different campaigns (for a gun control initiative, the Katrina Benefit, P&G's Tide Coldwater Challenge, and an advertising campaign that built in pass-along functionality). Each campaign reached more people than it would have by providing some type of technology platform that allowed people who initially experienced the message to share it with others.
OK, there's my summary. So, what's the news here as it applies to marketing communication?
First, the article is helpful to the extent that it reveals the limitations of viral marketing as an analogy -- companies and diseases are different. Companies can start their story with a lot of people, or a small group of people, while diseases typically start with one person and then spread to others.
Second, the article offers a metric -- the reproduction rate -- that can be used when quantifying and reporting the reach of a marketing program.
Third, the article presents interesting examples of companies using technology platforms that helps to amplify the spread of a message. One of these, which I would encourage people to learn more about, is ForwardTrack. This platform augments tell-a-friend functionality by incorporating geographic and social network tracking. The platform tracks the spread of messages from one person to the next and maps it, with the idea being that if a person can see the impact they are having on the spread of the campaign they will be more likely to increase their involvement. There are some important limitations to be aware of with ForwardTrack (for example, it tracks forwarding of messages only in an online environment, and not the content of what's said or other potentially rich details about the WOM episode), but it has a number of promising applications. I'll likely discuss more about this in a future post.
Why would this article or theory be helpful to organizations wanting to better understand marketing communication?
First, it shows how organizations can combine older and newer media to maximize the reach of their marketing initiatives. This is important because organizations shouldn't feel like they have to through the proverbial baby out with the bathwater.
Second, it provides one way for companies to measure the relative value of their marketing initiatives using quantitatively.
Third, it offers solid evidence that organizations, especially large companies with considerable resources, can limit their risk associated with a marketing initiative by starting with a large number of initial seeds.
What are some of the limitations of this article or theory?
First, the article seems to conflate two types of WOM marketing strategies: influencer marketing (starting with a small, select group of individuals with certain characteristics) and viral marketing (creating informative, entertaining, or otherwise engaging messages or experiences designed to be passed along to others, often via online means). See definitions provided by the Word of Mouth Marketing Association. According to these definitions, there's no reason why you can't use a viral marketing strategy and start with a lot of seeds. In fact, this is one way companies who want to use viral marketing strategies can reduce their risk -- seed the experience with more people. Further, the authors don't present evidence, at least in this paper, that starting with certain people who may be more well connected or have other characteristics (popularly known as "influencers"), might generate higher reproduction rates (though see Watts' work on "accidental influentials").
Second, I worry about the "uptake" of this article and theory by companies. Here's what I mean: depending on your philosophy of WOM marketing, "big seed marketing" could be considered a relatively "conservative" approach. That is, it seems to assume the best way to engage people is through a consistent message sent to large audiences. It would be an unfortunate result if companies came away thinking, "Phew, thank goodness! Now that I see these numbers I realize we can just proceed with business as usual but with a little twist: we'll keep creating advertisements, add some cool new tell-a-friend functionality, and we're set. We've now figured out this whole web 2.0, empowered consumer, networked world idea." Now, please don't misunderstand. I'm not implying that the authors of this article think this way or intended this. My point is that I think this is one prominent way this article will LIKELY be taken up and discussed in MANY companies. And if so, it would be a shame, because it misses out on the importance of more interactive ways of engagement between companies and their stakeholders: dialogue, listening, co-innovation, etc.
Third, and this is a more technical point, I question the value of reporting the reproduction rate as a final, or total value, rather than at each degree (see Table 1 and Table 2 in the article). Remember, the reproduction rate (R) is the number of new people converted based on existing people. So, rather than saying the reproduction rate for the whole campaign was x, I think it's more valuable to say the reproduction rate between the first degree seed to the second degree person was x. The better value to report as a total value is the "gain," or the multiple of new people reached (so, a "gain" of 2.0 means that if you started with 10,000 people, you reached 20,000). The authors do report the gain as a final value, but the confusing part is why the R is computed as a final value.
I corresponded with Professor Watts via e-mail about this point and learned the reason R was computed as a final value, rather than at each degree or generation, is due to the simplifying assumptions made in the paper (which, to their credit, the authors explicitly point out). One of the simplifying assumptions was that they assume the reproduction rate is constant across each degree of separation from the initial source (though, to their credit, they point out ways to make this more realistic). Further, they knew they were dealing with cases where the reproduction rate was under 1 at each generation anyway. But where this became confusing is that it makes it appear as though the formula for computing R is as follows:
Final R Value = "Bonus" Number of New People Reached/Total Reached (whereas a different formula for computing R [R = Bz ] was used earlier in the manuscript).That is, the number of additional people reached due to the pass-along/viral effect divided by the total number reached. When computing it this way the final R will always be less than 1 (because you can never reach an additional amount more than the total amount since the total amount is computed by adding the initial number of seeds and the number of new people gained). I would suggest that further research computes the reproduction rate at each distinct degree of separation from the source, and not as a final value. I'll have more to say on this point in a future post.
OK, in sum,I definitely recommend that people read this article, but hopefully keeping in mind what I see as some of the limitations. I'd love to hear what anyone else's thoughts are about the article and their theory.
Also, as a teaser, I'll be reporting some really interesting research soon that will be even more interesting when considered in light of the Watts et al. article. More to come!
Thanks to Professor Watts for his generosity in responding to my inquiries!
Tags: WOM word of mouth Word-of-Mouth Marketing buzz marketing viral marketing marketing communication big seed marketing