Category: Research Insight

Wharton Conference on User-Generated Content Part I

Posted on

In between the wedding and my race against the clock to get as much research done as possible before my research leave is over in January, the year 2009 has quietly slipped away and the holiday season is already upon us.  First of all, happy holidays!  As a gift to my readers, I want to bring some new exciting research insights from the conference The Emergence and Impact of User-Generated Content (UGC) I just attended in Philadelphia last week.  The conference was co-hosted by the Wharton Interactive Media Institute and the Marketing Science Institute, and featured top-notch researchers and practitioners who work in the field of social media and UGC.

A major question addressed by quite a few presentations at the conference was the impact of user-generated content. So in Part I of this two-part conference report series, I would like to highlight three presentations that I found particularly interesting with regard to this topic.

Philadelphia

Does consumer chatter about a product affect stock return?

The answer is yes, according to the research presented by Professor Gerard Tellis from the University of Southern California. In their research, Professor Tellis and his doctoral student Seshadri Tirunillai looked at six diverse product categories with rich consumer reviews: data storage, footwear, toys, personal computers, cellphones, and PDAs/smartphones. They gathered consumer reviews in these product categories from three sources: Amazon.com, Epinions.com, and Yahoo! Shopping. These reviews were then analyzed for the overall rating, review volume, and valence (positive or negative) of review associated with each product. Using a mathematical approach called vector autoregressive, the researchers tied these review characteristics to each company’s stock return and volatility. They found that consumer reviews lead stock performance by a few weeks (meaning that consumer reviews can help predict stock performance a few weeks ahead). Specifically, the volume of review (after controlling for the valence of review) has a positive effect on stock return.  The overall rating (e.g., 3.5 out of 5) did not have any significant impact on stock performance.  But the number of negative reviews and the average percent of negative expressions in the reviews negatively impact stock return and increase stock volatility. In contrast, positive reviews did not have a significant impact.

Lessons for marketers:

  • It is justifiable not only from a marketing perspective to monitor consumer opinions in social media but it makes financial sense as well. Research such as this can help make an argument to financial managers why a company should invest in such monitoring activities.
  • Although positive reviews may make one feel warm and fuzzy, it’s much more important to pay attention to negative reviews.  In general, negative information is much more diagnostic in conveying market sentiment.

Lessons for investors:

  • Consumer reviews may seem far removed from the complex mathematical modeling that goes into stock picking and performance prediction. But this research suggests the value for investors to monitor this social space.
  • The researchers further recommended a few investment approaches. For example, as a short-term strategy, buy a stock when its product review enters top 20% and sell the stock when it drops out of the top 20%. The recommended holding period for this strategy is 6 weeks.

Do bloggers affect product sales?

Bloggers like me probably would all like to know that we are making a real impact after the time and effort we’ve put into our blogs. Some companies also invest heavily in the blogosphere and want to know whether that’s a wise thing to do. The research presented by Professor Sriram Venkataraman from Emory University found that blogger influence is geographic-specific depending on the demographics of a market.  Using movie industry data, this research finds that a movie’s first-day national sales is not associated with blog variables.  However, when looking from the DMA (designated market area) level, strong geographic influence emerges. Not surprisingly, markets with a larger portion of young people are more likely to be affected by blogs and at the same time are more likely to discount the influence of company-sponsored advertising.  For markets with a higher proportion of female consumers, the research found that they tend to be more forgiving to negative blogs.  These consumers could read quite negative blogs about a movie but still feel and act positively toward the movie.

Lessons for marketers:

  • Consumer blogs can be a worthwhile tool to integrate into a company’s marketing strategy.
  • Selectively using these tools based on each market’s demographics may be more effective than a blanket strategy.

What about user contribution in new product development?

This research first struck me as using a very clever data source to address an important question.  Partially based on Professor Matthew O’Hern’s doctoral dissertation, this project uses the well-known open source community SourceForge.net to examine if user collaboration and contribution truly lead to better and faster product development. The answer is mixed. O’Hern and colleagues classified user contributions on SourceForget into three categories: (1) user reports: reports of bugs and issues found in a piece of software; (2) user requests: requests of new functionality or modifications to be added to future software releases; (3) user revisions: user-submitted solutions (i.e., codes) for fixing certain problems or adding new functionality to a software release.  They found that:

  • User reports of problems increase release activities, indicating a positive impact on software development.
  • At the same time, such problem reports alert other users of issues with the software and reduce the download volume for a software release.
  • User requests have the most negative impact, both reducing download volume and release activities.
  • Most surprising to me, users submitting their own solutions did not have any significant impact on release activities.  The only impact it had was on increasing download amount for a given month.

Lessons for marketers:

  • Wiki-type efforts by users may not always be beneficial to a company’s new product development.  When not properly managed, it can actually prolong the development process and reduce the speed-to-market.
  • Caveat: SourceForge is a community of mostly volunteers who do not have a strong commercial interest. Therefore, the proper utilization and integration of user revisions may be limited due to the lack of human power and resources. I would not be surprised that user submission will have a more positive impact in a more closely managed environment.

* * * * * * *

Plenty of information to digest for a while.  So I’m gonna stop here for Part I of the series.  What do you think of these research insights?  I’d love to hear back from you.  If you find any of these projects particularly interesting and would like more information, I encourage you to contact the presenter.  Whenever possible, I tried to provide a link to the presenter’s homepage so that you can find his/her contact information.

In Part II of this series, I will discuss another project on a privacy-friendly target advertising approach based on social network data.  I will also share with you a few high-priority topics related to social media and Internet marketing that were identified by practitioners at the conference.  So stay tuned!

Word-of-Mouth or Traditional Marketing?

Posted on

Some people may disagree with what I am about to say here: online social networks bring people closer to each other. At least that is the personal impact that they have had on me.  But what does this mean for marketing?  One answer is that word-of-mouth between consumers is carrying more weight in how we choose and consume products. Whether we love or hate a product, now it is so easy to make it known to the public that we are essentially affecting the opinions of other consumers (from total strangers to close friends) every day.

Managers are often hesitant to invest in encouraging word-of-mouth, however, as its effects are notoriously difficult to measure.  This is because word-of-mouth behavior is often unobserved, and it is difficult to tease out the concurrent impact of traditional marketing.  These are the exact problems a recent article by Michael Trusov and colleagues in Journal of Marketing tried to tackle.  Entitled “Effects of Word-of-Mouth Versus Traditional Marketing: Findings from
an Internet Social Networking Site”, this article offers a clear answer to the relative effectiveness of word-of-mouth vs. traditional PR and marketing.

Word of Mouth

What did they look at?
The impact of word-of-mouth, event marketing, and media appearance on the sign-ups for an undisclosed online social network.

Some intuitive findings:
More new sign-ups resulted in more word-of-mouth; event marketing led to more media appearance, and vice versa;  word-of-mouth was not affected by previous event marketing or media appearance, however, suggesting consumers’ relatively independent opinions and actions.

Some not-so-intuitive and very important findings:
The 3-day elasticity of sign-ups with respect to word-of-mouth was .17. In layman’s terms, this means that doubling the amount of word-of-mouth increases sign-ups by 17%. The corresponding impact from event marketing and media appearance, in contrast, was only 1.7% and 2.2%. The gap became even bigger with regard to long-term effects.  In the long run, the effect of word-of-mouth is 20 times that of event marketing and 30 times that of media appearance.  While doubling event marketing or media exposure led to 1.7% and 2.6% respective increase in sign-ups in the long run, doubling word-of-mouth increases sign-ups by a full 53%. Financially, an outbound word-of-mouth referral translates into 75 cents/year increase in advertising revenue.

What does this mean for marketing practice?
Word-of-mouth is a powerful tool for customer acquisition.  With the help of more powerful tracking tools provided by social networks and websites, it is possible for managers to measure the return from word-of-mouth activities. The mathematical approach used in this article (vector autoregressive modeling) further helps tease out the impact of other marketing and PR activities so that the true effect of word-of-mouth can be accurately measured. Together, this should reduce the hesitation to incorporate word-of-mouth into a company’s overall marketing strategy. The findings from this article also provide a strong motivation to better utilize word-of-mouth channel of communication.

Cautions
Readers should be cautioned from taking the results from the above research too literally.  Two things should especially be taken into consideration.  First, the data came from an online social network.  Customers on such websites are usually highly motivated to invite their friends, and those invited by their friends are also very likely to sign up.  If we were to change the context to, say, online banking, both the level of referral and the impact of referral are likely to be lower.  Second, the word-of-mouth activities studied in this article are all organic referrals initiated by consumers themselves. If the word-of-mouth had been stimulated by the company (say, with financial incentives), the referrals may not have been considered as genuine to other consumers and therefore may not have created as strong of an effect as reported in this study.  Although these are real limitations, the findings from this study are still quite powerful indicators of word-of-mouth effect. It is a tool managers should not ignore.

Reference
Michael Trusov, Randolph E. Bucklin, and Koen Pauwels (2009), “Effects of Word-of-Mouth Versus Traditional Marketing: Findings from an Internet Social Networking Site,” Journal of Marketing, Vol. 73 (September), p.90-102.

Are You Targeting the Right People to Grow Your Community?

Posted on

Last time I discussed a few research findings on what makes people pass on information to others.  This week, I would like to follow up on the topic and talk about a recent project done by Zsolt Katona (@UC Berkley) and his colleagues.  The research question Katona and colleagues set out to answer is what drives the growth of an online community. They surmised that the specific social network structure of the initial adopters affect the adoption likelihood of subsequent followers. To test their thinking, the researchers analyzed the first 3.5 years of data from a central-European social networking website, when no marketing activities had been engaged to promote the site and the network had been growing organically through word-of-mouth.  Here is the gist of what they found.

People do tend to follow the crowd but a more closely-knit crowd carries much more power

We all have hesitations when it comes to novel new things and may consider them risky. Depending on how risk averse we are, we may wait until some or a majority of other people have adopted the new thing before we jump onto the wagon. In my own research project documented in the last blog, we found the median adoption threshold to be 50%, incidentally supporting the “majority rules” mentality. But the threshold reported by our sample ranged across the whole spectrum from 0% to 100%. Consistent with this idea of an adoption threshold, Katona and colleagues found that more people in one’s social circle adopting a social network makes one more likely to join the network. In this context, perhaps an additional driver besides risk is the fact that the utility of a network increases when more of one’s friends belong to it. The story does not stop here, however. The researchers also found that a closely-knit (or high-density in network science terminology) network where everyone knows everyone else is much more influential. If the same number of individuals in a closely-knit network joins a social network, the remaining non-adopters are much more likely to follow suit than if it were a loose (low density) network of sorts.

Network

Social butterflies are not the most influential

In network science, the fact that some individuals have way more friends/connections than most others in the same network has often been compared to the rich get richer phenomenon.  But unlike the richer people who do have solid cash to spend, social butterflies who have tons of friends (think 1000+ or even 500+ Facebook friends) are actually quite weak when it comes to influencing other people’s opinions. Well, at least when it comes to the decision to join a social network any way. This may be surprising on first look. But not so when one thinks deeper about human psychology. We all have limited energy to build and maintain friendships. The more friends we accumulate on a regular basis, the less energy we have to develop a deep and meaningful relationship with each individual, and thus the less we are able to exert a strong influence.

Weak ties may be good for information travel but exert limited influence

The strength of the weak tie has been a well-known phenomenon for more than 20 years, referring to the fact that weak ties that link disconnected networks are critical to the spreading of information. However, for exactly the same reason, the central role played by these weak ties also makes a network formed around such ties more vulnerable.  Referring to these individuals as structural holes, Katona and colleagues found that the adoption of a social network by these structural holes has less of an impact on their friends, perhaps accurately reflecting the fact that these are “weak” ties.

Lessons learned

  • Many factors create counter effects when it comes to increasing awareness of a community vs. increasing participation in a community.
  • While sometimes it may be necessary to target loosely-knit networks (more weak ties) for increasing the awareness of your online community, closely knit networks are eventually critical to increasing actual participation in your community.
  • The same thing goes with highly-connected individuals. While those who have lots of friends may be good for getting the word out, individuals who have a more moderate friend circle may be more ideal for building the community.
  • For a business, how these counter effects should balance out will depend on the exact goal for the online community at each stage.

Reference

Zsolt Katona, Peter Pal Zubcsek, and Miklos Sarvary (2009), “Network Effects and Personal Influences: Diffusion of an Online Social Network“. The full paper can be downloaded from Katona’s website at http://www.cs.bme.hu/~zskatona/pdf/diff.pdf