Word of mouth has been guiding decision makers since the first Pleistocene hunter whispered a tip about a prime herd of mastodons. In the age of the Internet, word of mouth consists of online reviews, blog posts, tweets, and other electronic messages. Now, thanks to research by McCombs scholars and others, digitized word of mouth is providing marketers with important new insights into the power of this ancient channel.
In fact, according to McCombs marketing professor Leigh McAlister, analysis of online chatter is capable of providing insight about a company’s sales performance that until recently were unavailable between a firm’s quarterly financial reports. “It’s like X-ray vision,” McAlister says.
In a paper published in the July-August 2011 issue of Marketing Science, McAlister helped focus this marketing X-ray machine with the help of lead author Garrett Sonnier, an assistant professor of marketing at McCombs, and co-author Oliver Rutz, now a professor at the Foster School of Business at the University of Washington. Their findings, described in “A Dynamic Model of the Effect of Online Communications on Firm Sales,” indicated a strong connection between online chatter and a firm’s daily sales. Internet babble, it turns out, is valuable — savvy investors can use this insight to make money in the stock market.
This team is not the first to look into the effect of online messages on sales of products and services. Most prior research, however, has focused on the purchase-influencing ability of reviews and ratings, typically gathered from one or more websites. For this project, researchers used a Web crawler to troll a much larger swath of the Internet for any and all mentions of a firm and its products. Then, using proprietary sentiment analysis software, they categorized comments as positive, negative, or neutral.
The study uncovered compelling relationships between sales and chatter. Essentially, the results suggest that if online chatter goes up 10 percent, sales are likely to go up 10 percent, as well. One question is whether sales are driving chatter or the reverse. McAlister suspects a little of both: When more people buy a product, more of them make online comments, which drives still more purchases, and vice versa.
As striking as this finding is, it’s less dramatic than one McAlister, Sonnier, and McCombs statistics professor Tom Shively describe in a forthcoming paper that looked at the impact of online chatter on a firm’s stock price.
“It surprised the daylights out of me,” McAlister says of her latest analysis of chatter’s effect. “I had thought it might have something to with sales. I thought sales might have an impact on brand equity. And there might be a remote impact on firm value.”
But the relationship was stronger and faster moving than that. “It’s immediate,” she says. “The chatter changes and the stock price changes.” This sort of information is of interest to investors, and some hedge funds have already begun using online chatter to guide investment strategy.
It’s quite a jump from passing on shopping tips to moving stock prices, but the nature of online chatter makes it possible. Blog posts, forum messages, and the like are far easier to gather than 20th-century phone calls. And, because they are going from one communicator to many listeners, they are also more influential than conventional spoken and written communications — all factors that are attracting more and more researchers to the field.
Evaluating Fakes and Entropy
One is Dina Mayzlin, a Yale University marketing professor whose work McAlister cites in her most recent article. Mayzlin’s work looked at how firms can strategically manage consumer word of mouth.
In a 2004 Marketing Science paper titled, “Using Online Conversations to Study Word-of-Mouth Communication,” Mayzlin and co-author David Godes of the University of Maryland looked at the effect of Internet chatter on TV show ratings and found that conversations that were more dispersed were associated with higher future sales.
“We found that if the same volume of conversations were concentrated in one community, the show didn’t do as well as if the conversations were scattered across several communities,” Mayzlin says.
Pre-consumption Tweets and Movie Sales
Among the many interesting findings coming out of research into online chatter has been one by Huaxia Rui, a Ph.D. student in the Department of Information, Risk, and Operations Management at the University of Texas at Austin. In a paper included in the 2010 Proceedings of the International Conference on Information Systems, Rui and Texas colleagues Yizao Liu and Andrew B. Whinston looked the specific role of Twitter.
The result, “Chatter Matters: How Twitter Can Open The Black Box Of Online Word-Of-Mouth,” showed how tweets sent by people who had not yet seen a movie, which he termed “pre-consumption tweets,” can influence movie box office revenues similarly to tweets from people who had seen the film.
“One thing I found surprising was the pre-consumption word-of-mouth actually has a bigger impact than positive post-consumption word-of-mouth,” Rui says. “Usually you’d think if someone said something good about a product it would boost sales. But we found that if you say you want to buy something, even though you haven’t tried it yet, that has a bigger impact.”
Rui suspects one explanation is that many post-consumption Twitter posters aren’t planning to see a movie again, at least not soon. “For pre-consumption, it’s the opposite,” he says. “You’re very likely to consume that product in the near future.”
Future of Online Word-of-Mouth
For all the vistas of understanding opened up by the ability to analyze large volumes of influential online word of mouth, some major questions remain unaddressed, along with significant obstacles to their solution. For instance, Rui says one of the limitations of his research has been the way he divided tweeters into more and less influential categories by simply splitting them into those with fewer than 650 followers — the 90th percentile in his sample — and those with more. The difference is significant, he says, but to be optimal the division should be a continuum from least- to most-prolific tweeters.
Mayzlin notes that, while data from some social media channels are readily available, that’s not so with all of them. “You can get the data from Twitter,” she says. “Facebook is much harder to get the data.” If recent history is any indication of the near future, additional channels will continue to appear, and it remains to be seen how available data from them will be.
For McAlister, one limitation of her most recent paper is that it relied on daily sales data from just one firm, an unnamed online purveyor of durable goods. “It would be really nice to replicate this with more companies,” she says. A better understanding of online chatter is not just of interest to marketers interested in driving their own sales, she notes. One big issue raised by her research relates to competitive intelligence. “Are there people out there who are able to read my highly confidential information?” she asks. According to her work, the answer is yes.