Online product reviews have become ubiquitous — but does the text of these critiques tell a story that a star system can’t? How does a consumer’s definition of happiness affect buying habits? Are stock options always the best way to encourage risk-averse CEOs to invest in unpredictable but potentially value-creating projects? Can the desire for one reward be fulfilled with another? Professors Anindya Ghose, Cassie Mogilner, Christopher Armstrong and Jonah Berger, respectively, examined these issues — and what they mean for business — in recent research papers.
*[This article was first published by Knowledge@Wharton on October 26, 2011]
Beyond the Star System: Analyzing Text Reviews Leads to Better Pricing
While consumers who shop in brick-and-mortar stores typically are able to test and evaluate products before making a purchase, that is harder to do in the online shopping world. Indeed, online shoppers usually have to rely on word-of-mouth or user-generated product reviews to guide their buying decisions.
The common belief is that by looking at those product reviews — both the star rating system plus the number of reviews posted — one can predict how well a product will sell. The impact on sales of textual reviews — i.e., the comments that users include along with their star rating — is often ignored.
Ghose, a professor at the Stern School of Business at NYU and a visiting professor at Wharton, sees this as a waste of valuable information. He and colleagues Nikolay Archak and Panagiotis G. Ipeirotis, also at the Stern School, have published a paper in Management Science titled, “Deriving the Pricing Power of Product Features by Mining Consumer Reviews,” in which they quantify the value of the text comments in the user-generated reviews.
“The star rating system really doesn’t tell me much,” says Ghose. “I can look at 10 different brands and they all have star ratings of 4.5 or 5. That doesn’t indicate the differences in people’s opinions of each individual product or each feature, even though all that information — long verbal descriptions about what the users liked and didn’t like — is imbedded in their reviews.”
Researchers using text-mining techniques and other analysis have already come up with indicators of how positive or negative the reviews are. “But prior work in text mining does not reliably capture the pragmatic meaning of the customer evaluations; in particular, the existing approaches do not provide quantitative evaluations of product features,” the authors write. Their research, they add, teases out the economic impact of user-generated product reviews “by identifying the weight that consumers put on individual evaluations and product features, and estimating the overall impact of review text on sales.”
Ghose and his colleagues looked at the sales and reviews of digital cameras and camcorders on Amazon, focusing on those product features most frequently discussed by consumers (rather than the product descriptions provided by manufacturers). “With digital cameras, someone could talk about the screen size, the battery life, the megapixel resolution, the weight and so forth,” says Ghose. “What we do is quantify the value that each of these features has on final sales. The means to the end is to look at the effect on pricing power. As pricing changes, sales will change, too.”
According to Ghose, their work is applicable to all kinds of goods and services, ranging from hotels and clothes to restaurants and cars — “basically any product that has multidimensional attributes.” And it can be useful to everyone involved in online selling. For consumers considering a specific product, “our analysis parses out which features other users find most valuable, without those consumers having to tediously go through hundreds of reviews. This reduces the cognitive costs for them at a time when social media is exploding.”
As for retailers and manufacturers, he adds, “we can tell them, going forward, which features they should be emphasizing in their next product design and where, for example, they should be putting more of their R&D. For advertisers who have limited space in virtual real estate, we can tell them which features they should advertise and highlight the most.”
The researchers’ paper also offers insights to search engine advertisers who rely on customer-generated opinions that automatically create an online advertising strategy using the sponsored search advertising model. For instance, the authors note, “our methods can be [extended] to different product categories [where] firms [can] select the appropriate keywords to bid in these advertising auctions that highlight the most pertinent differentiating characteristics that consumers value.”
Ghose offers an example. Suppose, for a given product, that the phrase “excellent video quality” is associated with an increase in sales three times greater than the phrase “great design.” The retailer or manufacturer would clearly be wise to choose a set of keywords that are associated with the first phrase rather than the second, and bid more on those keywords in paid search advertising.
Amazon, according to Ghose, is aware of their work, as are several startups. If Amazon were to implement their approach, Ghose says, it could be in the form of a drop down menu for the different product features, letting consumers choose the ones they value the most. The site would then summarize the best options available.
“We have shown how textual data can be used to learn consumers’ relative preferences for different product features and also how text can be used for predictive modeling of future changes in sales,” Ghose says. Increasingly, he adds, “we are beginning to see that social media content — such as ratings, reviews, blogs, tweets, social networks, etc. — can have a business impact that is being reflected in economic variables (for example, product sales, pricing premiums, profits)…. Our methods combining automated text mining on unstructured data with statistical and econometric modeling can help businesses precisely quantify the dollar value of social media on their sales and pricing power.”
How Happiness Resonates in the Check-out Line
The dictionary defines happiness as a “state of well-being and contentment.” In reality, however, people feel and experience happiness in a variety of different ways and for myriad reasons.
In two recent papers, “The Shifting Meaning of Happiness,” and “How Happiness Impacts Choice,” Wharton marketing professor Mogilner, Stanford University professor Jennifer Aaker and MIT professor Sepandar Kamvar conclude that the meaning of happiness changes based on a person’s temporal focus: Those who are focused on the future are more likely to associate happiness with excitement; when a person is concentrating on the present, however, happiness becomes more closely linked to a feeling of calm.
As marketers try to establish a more emotional connection with customers, and recession-weary consumers seek products that provide brief moments of pleasure, a clearer definition of happiness is becoming more important to the business world, Mogilner says. The paper notes that brands — including Dunkin Donuts, Nivea, Hugo Boss, Clinique and Coca-Cola — have launched campaigns that connect their products with happiness. (In fact, Clinique sells a perfume that is simply called “Happy”.)
“People are incredibly stressed out because of the economy, [and because of] losing a ton of money,” Mogilner says. “They’re realizing that what really matters at the end of the day are the simple things, namely being happy…. Instead of selling luxury, marketers are trying to connect on that more simplistic level.”
For “The Shifting Meaning of Happiness,” which was published in Social Psychological and Personality Science, the researchers studied sentences expressing emotion posted on 12 million personal blogs to see what other words were mentioned when the bloggers said they felt happy. They also conducted a series of experiments with adults in a broad range of age groups. The results pointed to a shift in the meaning of happiness as people age, from a state associated with excitement to one equated with peacefulness.
The results of experiments from the second paper (forthcoming in the Journal of Consumer Research), “How Happiness Impacts Choice,” showed that “the meaning of happiness is malleable, shifting both moment-to-moment and over the course of one’s life.” Although people naturally redirect their attention from the future to the present as they get older, a similar change can occur based on someone’s actions and environment, Mogilner notes.
A customer’s mindset at any given time plays a role in what products he or she chooses to buy. For example, the researchers recruited 50 adults ages 21-49 to participate in a new product survey that asked them to choose between two package designs for the fictitious bottled “Happiness Water” and to rate how happy they thought the product would make them feel. One package design used a “soothing” green-colored droplet and the tagline “Pure Calm.” The other depicted the same droplet, this time colored bright orange and including the description “Pure Excitement.” “The reason we used water is that it is clearly the most neutral product out there, so branding is what really differentiates consumers’ options,” Mogilner says.
Before being asked to choose a logo, the respondents completed a sentence-unscrambling task designed to direct their temporal focus toward the future or the present. Participants prompted to think about the future by the sentence task were slightly more likely (54% to 46%) to choose the “exciting” product over the calming one. Those influenced to concentrate on the present overwhelmingly (78% to 22%) chose the “calming” product over the “exciting” option.
Keeping this effect in mind, marketers can adjust their strategies based on the demographic they are trying to attract, or a consumer’s anticipated state of mind when encountering a product, according to Mogilner. “You would think that someone’s definition of happiness would be very fundamental to their values or philosophy or personality. But just a simple meditation exercise or even being exposed to future-related or present-related words can shift someone’s mindset so dramatically as to influence how they define happiness and the choices they will make to obtain it.”
CEO Stock Options and the Risk vs. Value Tradeoff
Imagine that the CEO of an innovative high-tech firm is sitting in Silicon Valley right now and debating whether to invest in one of the project proposals on her desk. According to the chief executive’s analysis, some of the projects have a positive net present value while others don’t, and all carry some risk that could affect the firm’s share price. Which one will she choose?
In their paper, “Executive Stock Options, Differential Risk-Taking Incentives and Firm Value,” Armstrong, a Wharton accounting professor, and doctoral student Rahul Vashishtha suggests that the firm’s shareholders might be surprised by the answer. The shareholders know their CEO is risk-averse, as most CEOs are, but they believe the firm has provided her with enough incentives — notably in the form of stock options — to encourage her to go for the project adding the most value, even though it carries idiosyncratic risk that will affect the company or, at most, the tech sector. Instead, according to the research, the CEO might opt for a project adding less value because the risk it carries is systematic and won’t just affect the company, but a much wider range of asset classes. “Contrary to conventional wisdom, if you want to encourage a CEO to take on idiosyncratic risk, stock options might not always be the best way to do it,” says Armstrong.
Yet stock options are a huge component of aggregate CEO pay in the U.S. The Wall Street Journal/Hay Group 2010 CEO Compensation Study published in May indicates that total direct compensation of chief executives at the U.S.’s largest companies averaged more than $9.2 million. The value of stock options and other long-term incentives of their pay packages accounted for $6.2 million of that figure.
With that in mind, Armstrong and Vashishtha used a research design that takes into account the non-random nature of the relationship between equity incentives and the types of risk to which companies are exposed. They then began sifting through a database of the proxy filings between 1992 and 2007 of 13,223 large, publicly traded firms in the U.S. The sample data set allowed them to analyze each CEO’s stock and stock option holdings over the course of his or her entire tenure at a company. “If a CEO has been around for six years, he has six years worth of grants,” notes Armstrong. “He probably hasn’t sold too much, and the value of his holdings is very sensitive to movements in stock price.”
He says for typical CEOs in the sample, every 1% change in their companies’ stock price led to their stock and option portfolio changing by $718,000 on average. “That’s a big swing in value, and [CEOs] have pretty big incentives to increase a stock’s value by, say, 10%, which would increase a CEO’s holdings by $7.2 million on average,” he notes.
Armstrong points out that CEOs tend to be more risk-averse than shareholders because they are under-diversified. “Most of their wealth is tied up in their company” in the form of stock and option holdings and their employment is tied to their company. “So if the company goes bankrupt, they lose their jobs and salaries, and their reputations are tainted.” In contrast, shareholders are presumed to hold a diversified portfolio, “which is why they generally only care about systematic, rather than idiosyncratic, risks that can knock some or all of their portfolios off course.”
According to Armstrong and Vashishtha, factoring in the knowledge that not all risk is created equal is one new way their research contributes to the growing body of academic work in the field. However, their research also explores the interplay between the two types of risk and the sensitivity of stock option value to return volatility (due to their limited downside, or convexity) and the sensitivity of both stock and stock options to share price.
First, they found that CEOs whose stock and stock option holdings are more sensitive to return volatility have incentives to take systematic, but not idiosyncratic, risk. “That leads a CEO to want to choose projects that have systematic risks over those with idiosyncratic risks,” according to Armstrong, because CEOs can hedge away any unwanted systematic risks, while idiosyncratic risks by their nature can’t be hedged.
Second — and arguably more important, says Armstrong — is what they found out about CEOs’ incentives to increase stock price. “We found a positive association between CEOs’ incentives to increase stock price and both systematic and idiosyncratic risk,” he says. In other words, “the limited downside, or convexity, that is unique to stock options relative to stock only gives CEOs incentives to take systematic risks, but the increase in value of both stock and options as share price increases, gives CEOs incentives to take on both systematic and idiosyncratic risk.”
According to Armstrong, chasing after systematic risks could be a waste of time and money, and if many CEOs do so, the investor community could face an increasingly homogenous marketplace and generally lower firm values. “We make the point that if there are a lot of options granted in the economy, all the CEOs might be searching for the same systematic risk, and may make the returns across the economy more correlated than they would [otherwise] be, making it harder for shareholders to diversify,” he says. “I could see that as a byproduct.”
Can Hunger Make People Prefer Distinctive Products?
Conventional wisdom would say that the only thing that can satisfy hunger is food. While this may be true in terms of the stomach, the brain is another story.
In a recent research paper, “Food, Sex and the Hunger for Distinction,” Wharton marketing professor Berger and Stanford University professor Baba Shiv found that fulfilling the desire among many people to distinguish themselves from others can stand in for a more basic need, such as hunger or sex. This tendency, they argue, has implications for any retailer hoping to better understand why consumers buy certain products. “Marketers need to think about how atmosphere and the environment might affect a shopper’s motives,” Berger says. “We’ve all been to the mall and walked by Cinnabon or Auntie Anne’s pretzels. The aroma definitely makes us hungry, but it might also make us hungry for products that set us apart from others.”
The researchers conducted four studies to see if the need for distinctiveness impacted the desirability of other, unrelated rewards. In one experiment, participants were approached outside a college campus dining hall and asked to choose their preferred options in five different product categories, including cars, printers and television sets. Each survey that was distributed had already been completed by a (fictional) prior participant, so respondents made their selections with the knowledge of what someone else had chosen.
They found that participants who filled out the survey before dinner were more likely to pick products that hadn’t been selected by the prior respondent — in other words, to make the more distinctive choice. “Neurologically, these rewards may be swappable,” Berger notes. “When you desire one type of reward, other types of rewards become more desirable…. It’s not going to make the hunger disappear, but it’s satiating your general reward system to some degree.”
In another study, the researchers showed male participants either “sexually arousing” images, such as swimsuit models, or control images. Then, the group was asked to take a survey about making choices among products with varying levels of uniqueness. But half the participants were given a surprise reward (a candy bar) intended to satiate their arousal before completing the survey, while the other half got nothing. The researchers found that participants who had not been satiated with candy were more likely to pick distinctive products.
According to the paper, the findings can be used by companies deciding where to locate a store or how to position their marketing campaigns. “Products related to distinctiveness (e.g. wine) might benefit from being placed near appetizing food samples, for example, and brands positioned around distinctiveness might want to locate near the mall food court, as they might benefit from hunger-inducing smells.” In addition, Berger says the research presents evidence of a need to slightly alter the old adage that “sex sells.” He argues that it does, but only in the case of a brand like Gucci or BMW, which position themselves as unique.
While the above suggestions might prove successful in the United States — where distinctiveness is a virtue — brands doing business in countries where less value is placed on being unique, such as many countries in Asia, may have to take the opposite strategy. “Americans prefer to think of themselves as different. We like to think that we are special unique snowflakes. In Asian cultures, what is valued is being a part of a community,” Berger notes. “In that cultural context, you might want to locate more conforming brands near restaurants and avoid locating distinctive brands there. Those fundamental drives are still affecting people’s preferences, but in this case, it moves them in the opposite direction.”
*[This article was first published by Knowledge@Wharton on October 26, 2011]
The views expressed in this article are the authors’ own and do not necessarily reflect Fair Observer’s editorial policy.
For more than 10 years, Fair Observer has been free, fair and independent. No billionaire owns us, no advertisers control us. We are a reader-supported nonprofit. Unlike many other publications, we keep our content free for readers regardless of where they live or whether they can afford to pay. We have no paywalls and no ads.
In the post-truth era of fake news, echo chambers and filter bubbles, we publish a plurality of perspectives from around the world. Anyone can publish with us, but everyone goes through a rigorous editorial process. So, you get fact-checked, well-reasoned content instead of noise.
We publish 2,500+ voices from 90+ countries. We also conduct education and training programs on subjects ranging from digital media and journalism to writing and critical thinking. This doesn’t come cheap. Servers, editors, trainers and web developers cost money. Please consider supporting us on a regular basis as a recurring donor or a sustaining member.