Data Mining vs. Science: How OkCupid translates data into dates

By Anne Mahoney

If you’ve ever marketed yourself online in the romance department, you likely have painstakingly analyzed every part of your profile. Is the photo attractive? Do I say the right things about myself? How should I word my first email to another enticing product in the dating aisle?

An article in the New York Times looks into how OkCupid, an online dating site, shares user data with its registered love-seekers to provide advice on how to develop and market their personal brands. To find the data, it analyzed 7,000 user profiles, noting photos, number of responses and content of those responses. One useful finding was that being “fascinating” or “cool” is more important than the initial physical attraction factor. For instance, OkCupid says a woman using a photo portraying her playing an instrument or on an exotic beach receives more responses than focusing on physical assets.

If true, this certainly is valuable information for site-users. The study has, however, strictly focused on pure numbers through data mining. To gain additional insights, I sought out the opinion of the foremost expert I know in statistical data analytics: Medill IMC professor Edward Malthouse. Professor Malthouse brought up the scientific question still at large for OkCupid: Why do these tactics work? He thinks marketing can help to explain.

“Some physical beauty is a point of parity,” he said. “Differentiators will make you stand out, at least among a segment that values such activities. So, marketing theory predicts that those who are differentiated will be more successful.”

That is the scientific way to view the findings of OkCupid’s study. It hypothesized that differentiators would increase or decrease response rates, which the study confirmed. Yet there are other variables that have not been taken into account. Malthouse points out an example as the experiences of the customer, or date-seeker browsing through profiles. These experiences are not directly focused on the “product,” or person trying to find a date. If each individual created a first-impression experience for the type of person they are trying to attract – the “targeted consumer” – Malthouse theorizes it would have an even stronger effect.

Boiling down an intimate subject with numerous and oftentimes-mysterious factors, such as dating, into pure numbers surely has its challenges. Do you think there’s truth to the OkCupid study? How can marketers benefit from activating these types of analytical tactics?

***

Anne Mahoney is the Social Media Director at Vitamin IMC and a student in the Masters in Integrated Marketing Communications program at Northwestern University's Medill School. She can be reached at annemahoney2010@u.northwestern.edu

Trading privacy for data

According to Stephen Baker, the Numerati are taking over. These mathematicians and computer scientists are the focus of Baker’s book, The Numerati, which he discussed last week at Medill. Baker described how our observed behavioral patterns are being turned into quantifiable data by these innovators, and hinted at the vast implications this has for marketers.

“This science is ideal for those industries where you can afford to make a lot of mistakes,” Baker said. That is, marketing and advertising. In practice marketing is a blend of art and science, intuition and analysis. The Numerati can analyze the vast amounts of data and uncover new insights about consumer tastes, preferences, and moods. The use of behavioral data undoubtedly makes marketing more effective than traditional segmentation approaches based on demographics.

Forget about demographic and psychographic targeting. NetFlix and Amazon.com use vast databases that predict behavior and make recommendations based on past purchase behavior. Sometimes the suggestions are misguided, but there is little cost associated with the error. It is certainly more accurate than a strategy based on customers’ billing zip codes.

But these new methods aren’t without problems. People are increasingly protective of their privacy and anonymity, both online and offline. Baker argued that consumers would be more willing to share information if they understand how it benefits them.

The ultimate effects of the Numerati’s efforts in using the staggering volume of available data will be decided by the public’s willingness to trade its privacy for better relationships with brands and businesses – or the government’s eagerness to step in and change laws.

Do you think companies are going too far with your private information?

--Kelly Kross

***

Kelly Kross is a graduate student in the Masters in Integrated Marketing Communications program at Northwestern University’s Medill School. She loves her BlackBerry and will never switch to the iPhone. Kelly can be reached at kellykross2010@u.northwestern.edu

AmEx's Data Mining Techniques Spook Its Customers

One distinguishing feature of IMC is its strong emphasis on data analytics. Now more than ever, quantitative insight and analytical skills are essential tools for gleaning meaningful consumer insights from the overwhelming surfeit of consumer data that most large corporations possess.

As both a marketer and a consumer, I find these treasure troves of data both exciting and somewhat terrifying. Last week at Davos, Richard Edelman bemoaned the general erosion of trust in corporate entities, and to be sure, this is a critical issue for IMC as well. Analytical skills are important, but being able to build relationships with consumers based on mutual trust is what ultimately builds shareholder value over the long term.

In this volatile economic climate, then, consider this New York Times article from January 30 that revealed American Express’ targeted rate increases for cardholders who used their Amex cards at specific establishments:

The question, then, is how much of the data they can use before spooking their customers. Kevin D. Johnson, a 29-year-old Atlanta resident who runs a marketing and communications firm, received a letter from American Express last October saying that his credit limit was being lowered. One reason was that other customers who had used their cards at places where he had shopped were late in paying their bills.

Read the full article at the New York Times.

UPDATE: After Good Morning America picked up this story, American Express clearly felt the sting of the bad P.R. of this data mining technique, and announced that it would stop the practice.

Does this practice take regression too far? Is there a line between “choosing your best customers” and outright discrimination? Would this policy incite you to choose a different credit card company?

We'll keep an eye out for future stories about issues in marketing ethics. And send us your suggestions!

-- Colleen Maley