Marketing and Data Science

Data, Data and Even More Data: Harvesting Insights from the Data Jungle

Bernd Skiera

Mathematical skills are a precondition for generating insights from data, but marketers need to react to diverse challenges to improve customer value.

From data wasteland to data jungle
Increasing global digitalization brings huge and ever-growing amounts of data. It all began with the invention of the browser that made access to the Internet via a desktop computer so much easier and faster. More and more consumers started to enjoy direct online interaction with each other and with companies… and started to leave their traces. The cost of observing these interactions fell to marginal costs that were very close to zero. It was, for example, possible for the first time to observe on a large scale not only that an advertisement was shown to a consumer, but also how the user reacted to that ad. So marketers were able to measure whether the consumer clicked and even purchased after clicking on the ad. Previously, a comparable measurement of advertising success was only possible for direct marketing activities but the cost of doing so was much higher and the quality of measurement much lower. For example, direct marketers could not even observe whether the consumer opened the letter they sent. Just compare this opportunity to the ones that email marketers have today.

The next major step forward came with the availability of affordable and powerful smartphones and mobile data plans. They now enable companies to target consumers everywhere, add location-based information to consumers’ actions, and record consumers’ reactions at the location and the time where the reaction occurs. Thus, instead of interacting with consumers during the few hours per day that they use their desktop, companies can nowadays interact with consumers essentially 24/7. The availability of data exploded and Hal Varian, chief economist at Google and previously a well-known researcher in microeconomics, became famous for saying around 2005 that “the sexiest job in the next 10 years will be statisticians.” So, instead of a data wasteland we seem to be living in a data jungle full of ripe fruit. Can marketers simply pick it up now? Is all of it wholesome? Or is harvesting insights from a data jungle a more challenging task than anticipated and one that requires new skills?

What companies can gain from big data analysis

  • Insights from academic research
    Many companies are convinced that the fruit of the data jungle is wholesome. Insights that arise from big data analyses are in high demand. In contrast to ten years ago, the number of company jobs for PhDs in marketing and economics is growing and growing. Well executed academic studies attract huge interest among companies. Managers are, for instance, willing to use the insight of a study by the researchers Blake, Nosko and Tardelis in 2015 on the unprofitability of Google AdWords for branded keywords to reallocate millions of dollars of advertising budget to other advertising media. The Wharton Customer Analytics Initiative (WCAI) successfully draws the interest of Fortune 500 companies to sponsor competitions that attract academics to analyze the data of those companies and share the respective insights. The list of marketing problems that are analyzed on the analytics platform Kaggle is constantly increasing, and online classes on “machine learning” are among the most popular online courses.
  • Improving marketing decisions
    Marketing can be much more effective if more and better information is available. With this special issue, we intend to help companies participate in the gains in efficiencies. I am glad that some of the most prestigious researchers in marketing took their time to contribute and highlight how marketing decisions can be improved in our data-intensive environment. Mike Hanssens, Pradeep Chintagunta and John Hauser illustrate how data can be transformed into useful information for various marketing applications. Recently, they served as co-editors of a Special Issue on big data of the flagship journal Marketing Science, and they share some highlights with us. In their article they also predict a much stronger collaboration of marketers with data scientists and computer scientists and stress that marketers can not only benefit but also contribute by modeling structure and exploring causal relationships.

    Martin Spann shows how to get location data, how to analyze it and how to use it to make customized marketing decisions. Including the situational variables that smartphone applications generate helps companies to design their offers so that consumers perceive them to be much more relevant. Martin Spann has a strong background in marketing and now conducts much of his research into information systems. Other researchers came the opposite way. These developments in research interests are fine examples of how the two fields, marketing and information systems, are moving closer together.

    Martin Schmidberger looks at the data jungle from a managerial perspective. For many years he has been the head of data and marketing at ING DiBa. Remarkably, this is the only bank whose market share in the German retail banking market has grown substantially grew over the last ten years. In my interview with him, he confirms that systematic use of data and machine learning techniques leads to a better understanding of customer behavior. Their systems generate customized recommendations with a better response than traditional marketing achieved and thus highlight again the fact that data science and marketing are a winning team.


Marketing challenges for harvesting insights from the data jungle
However, fruit-picking in a data jungle is not automatically rewarding. It can be very tricky, sometimes even painful to harvest true insights. Mathematical skills are a precondition (see Box 1), but marketers need to react to diverse challenges and our authors deal with other critical issues as well.

  • Drawing correct conclusions
    Two of the strongest women in our field, Anja Lambrecht and Catherine Tucker, highlight the important difference between correlation and causality. On the one hand, this topic is an old one and every student hopefully learns that a regression analysis indicates correlation but not necessarily causation. On the other hand, the ability in a digital world to target individual consumers makes this problem so much more important. In Box 2 you find an example that illustrates how tempting it is to draw wrong conclusion. Hopefully you will see that only experiments like the ones Anja Lambrecht and Catherine Tucker describe are able to observe the causal effect of marketing actions and that only those causal effects should guide marketing allocation decisions.


Bernd Skiera, Profesor of Electronic Commerce and E-Finance Lab, Faculty of Business and Economics, Goethe-University, Frankfurt am Main, Gernany

Further Reading

Blake, Thomas, Chris Nosko and Steven Tadelis (2015), “Consumer Heterogeneity and Paid Search Effectiveness: A Large-Scale Field Experiment”, Econometrica, Vol. 83, No. 1, 155 – 174.
Chen, Andrew (2012), “Growth Hacker is the new VP Marketing”, http://andrewchen.co/how-to-be-a-growth-hacker-an-airbnbcraigslist-case-....