Learning From the Dynamics of Viral Marketing

January 25, 2010

In what context do viral marketing strategies work? How do on-line product recommendations develop, multiply, spread and ultimately, dwindle and die? And, can Marketers influence any of this?  

These were important questions posed by Leskovec, Adamic, and Huberman in their 2008 study “The Dynamics of Viral Marketing.” This is one of the few studies I’ve seen to actually study how on-line recommendations grow virally and how this growth impacts purchase behavior throughout the viral network.  

Viral Marketing -- When and Where Does it Work?

 

Admittedly, the study had limitations, notably that it was only four categories, just measured on-line viral activity, and included a discount incentive to help motivate purchase. But, even with these limitations, it uncovered deeper insights into the systematic patterns in knowledge sharing and persuasion online—all of which are of high interest to Marketers.  

What Did the Study Entail?

Lescovec et al. examined an online recommendation network composed of 4 million people who made 16 million recommendations for 0.5 million products. Each time a consumer purchased a book, music, or movie, he or she could  send e-mails recommending the item to friends. The first person to purchase the same item through a referral link received a 10% discount.  

What was Measured

  • When and at what price a product was purchased
  • If the product was recommended to others
  • Whether the recommendation resulted in a subsequent purchase and discount

They then modeled the effectiveness of recommendations as a function of the total number of previously exchanged recommendations.  

Recommendation Networks Grow Slowly Over Time.

 

Important Viral Network Learnings & Insights

Finding #1:  Consumers recommended a large number of products to the same group of people. As a result, recommendation networks became heavily locally-based. For example, in the DVD recommendation network there are 182,000 pairs that exchanged more than 10 recommendations.  

Consumers Tend to Recommend Products to the Same People

 

Finding #2: Recommendation networks centered on a specific product category. That is, the people tended to focus on recommending a particular product category and thus created a “community of interest.” Having said this, most all networks shared recommendations for all types of products.  

Finding #3: Trust, influence, and perception of “spam” affected purchase. As people exchanged more recommendations, the likelihood they would purchase the product increased due to a growing foundation of trust. However, purchase likelihood increased, peaked, and then fell as consumers received additional recommendations for a specific product. A few recommendations built credibility; too many appeared as “spam.”  

Finding #4:  Most recommendation chains didn’t grow very large. In fact, most terminated with the initial product purchase, and even the largest connected networks were very small as a percentage of the total population.  

Recommendation Chains Don’t Typically Grow Very Large

 

Finding #5: 20% of recommendations accounted for 50% of sales. This is not far from the usual 80-20 rule, where the top 20% of products account for 80% of sales.  

What are the ‘Viral’ Implications for Marketers?

1. Identify the “Amplifiers.” Given that 20% of recommendations generate 50% of sales, it’s key to figure out whom the amplifiers are and focus your efforts on them.  

2. Determine Where the “Amplifiers” Congregate. Where do they exchange product information? On what platforms do they consume media? Web behavior can be linked to off-line purchase panels to quantify the effectiveness of recommendations (see “What Really Drives Web Advertising ROI”).  

3. Take Online Recommendation Networks to the next level Through Social Media Marketing. Marketers should explore development of models to measure recommendation systems on Twitter, Facebook, Foursquare, and the larger online arena. Through broader web 2.0 outreach, marketers can quantify consumer engagement on recommendation networks by volume, reach, tone, and source.  

Marketers can optimize paid media and earned media with viral marketing.

 

4. Be Wary of Creating “Recommendation-Fatigue.” A fine line exists between trust and influence in recommending a product and what is widely considered “spam.” Consumer engagement via any online channel must be done with careful consideration of earned media and buzz promotion.  

Viral Marketing: Limitations…

What’s not yet so clear from the research is how to minimize transmission “breakdown” – e.g. how do you minimize the likelihood that a product recommendation is the last one. As the research showed, most viral networks don’t grow very large. Marketers will only invest significant money if they can truly scale viral marketing programs.  

…And Future Opportunities

With the right tools and metrics, marketers can diversify their marketing plans to incorporate viral marketing strategies. The research clearly shows that viral marketing can build unique and niche recommendation networks, bolster consumer engagement, and lift sales.  

And as consumers continue to favor a digitally-based, social network-centric world, it’s critical that Marketers become more expert at viral marketing. Key to this will be identifying amplifiers, focusing on congregation points, leveraging social media opportunities—all without overdoing it. As importantly, Marketers must discover new approaches to spread and scale viral marketing just as effectively as the flu seems to proliferate every flu season.  

Follow Randall Beard on Twitter  

Get free updates of Randall Beard’s Blog by e-mail  

Get free updates of Randall Beard’s Blog by RSS reader