We have a suspicion: Is it possible that our expectations of marketing using big data cannot be fulfilled? Because there is simply too little data in our country compared to the USA, where most of the success stories that point the way here also come from? It is difficult to provide evidence for this conjecture because much is a matter of judgment and there is a lack of well-founded insights. Therefore, we rely on the following circumstantial evidence.
First, let's take a look at the USA, which lives up to its reputation as the land of opportunity in these matters. The Federal Trade Commission (FTC), which published a report on data brokers in May of this year after almost two years of investigations, provides good insights here. Nine selected companies (Acxiom, Corelogic, Datalogix, eBureau, ID Analytics, Intelius, PeekYou, Rapleaf and Recorded Future) were asked to answer an extensive list of questions. Most of these companies offer products for marketing, while others are active in the field of risk mitigation or tracing people. Because, unlike in Europe, legislation in this area is not based on a fundamental right, but comes into play when concrete damage has occurred, the handling of data is much freer.
For many companies, the obvious thing to do is to sell their data for as much profit as possible. In negotiations between business partners, it is also common to regulate the exchange of data. At the center of these activities is the complex and correspondingly confusing industry of data traders. Data is not only supplied to end users via several trading levels, but is also exchanged between these brokers. These brokers themselves collect data at all levels of government, such as bankruptcy information or voter registrations, and from commercial companies that provide data on consumer purchasing behavior, web behavior, warranty registrations and other data on everyday consumer behavior. More than half of the data brokers surveyed stated that they also analyze other publicly available information, such as telephone and other directories, newspaper articles and information that consumers publish on the Internet, such as on blogs and in social media.
It so happens that data brokers store billions of information elements, capturing virtually every American consumer. One of the new data brokers surveyed stored 1.4 billion pieces of information about business transactions with consumers, creating 700 billion aggregated data elements. Another data broker's database includes $1 trillion worth of transactions with consumers. And yet another data broker adds three billion new records to its database every month. So it's not surprising that one of these data brokers has 3000 data segments for just about every American consumer. The business of data brokers is not easy: a sophisticated technical infrastructure has to be operated, data not only has to be procured but also checked for quality and, if necessary, additionally processed and updated again and again. And then it is a matter of drawing the right conclusions from all these mountains of data by means of statistical analyses. Where "right" here means classifying consumers into more and better categories and achieving higher hit probabilities than the competition.
"Mobile Mixers" and "Married Fine Spirits"
Harmless is the assignment to consumer interests. For example, "dog owner," "winter sports enthusiast," or "responds to mail-order advertising. It becomes more critical when consumers are categorized by ethnicity or income: For example, low-income Latinos or African-Americans in particular are categorized as "urban mix" or "mobile mixer." Other tricky specifications concern, for example, the age of the consumers. "Rural undecayed" includes single men and women over age 66 who have low levels of education and wealth. Meanwhile, "married sophisticates" include childless upper-middle-class couples in their thirties. Other delicate categories involve aspects of health: "expecting parents," "interest in diabetes," and "attentiveness to cholesterol."
What makes European data protectionists' hair stand on end is the basis for good business for American data traders: data from the online and offline worlds are combined. This is done primarily using information that is entered when registering on websites, especially the users' postal address. The web users' browsers are then marked with a cookie so that they can be shown advertising content based on their offline activities as well.
Everyone squats on their own data
This American practice stands in stark contrast to the Swiss reality: in the survey conducted by Werbewoche on big data at large Swiss companies (cf. Advertising Week 19/2014 of 24.10.2014, p. 36f) always sounded the same when asked about the disclosure of data, including anonymized data. A typical example: "Swisscom complies with the legal requirements at all times and protects the privacy of its customers. In particular, Swisscom does not sell any personal data to third parties. Such a procedure is excluded by the Data Protection and Telecommunications Act and our GTCs. When processing location data, there are clear legal requirements in Art. 45b of the Telecommunications Act, which we naturally comply with. Commercial use of personal location data would therefore only be possible with the customer's consent." However, the telecommunications giant is also the only example in the survey that reported the passing on of anonymized information: "This is sold on the market - for example, anonymized average speeds and travel times on freeways as part of the 'Travel times' project for the Federal Roads Office (FEDRO). " The situation is different for the other companies inquired about, such as the two major distributors: "Coop does not pass on anonymized or highly aggregated data to third parties either." And: "We do not make any data available outside the Migros Group." The suspicion is confirmed: if no one passes on data, everyone is simply relying on their own data, which may be supplemented with generally available data on demographics, etc.
It becomes clear that if you can only evaluate your own data with Big Data analytics methods, you are, to a certain extent, sitting in a data desert and cannot swim in a data lake. Then, even with the most modern technology and the best personnel, you won't get anywhere, just as when broadly collected and diversely processed data can be used according to the American model. For example, a (fictitious) major retailer sees that customer Anita always shops at the same store, but never has any meat in her shopping cart. So Anita is addressed as a vegetarian and pointed to the latest vegi products. But Anita buys her meat needs from the butcher she trusts, especially since her big-box store doesn't have an open meat counter. If the wholesaler could (as in the USA) compare its data with the findings of the data brokers (where, for example, information from payment transactions is included), it would align its communication with Anita in such a way that she would also buy meat products from it in the future - instead of irritating her with grain picker information (and damaging its own brand in the process).
No sign of a data protection U-turn
In order to achieve better framework conditions for the use of big data in marketing, there needs to be a veritable about-face in data protection legislation and ultimately a Swiss go-it-alone approach to data protection. This will only succeed if there are industry-wide coordinated procedures to strengthen the public's trust in commercial data collection and thus pave the way for politics. However, there are no signs of such an initiative: In the Werbewoche survey, no agreement was found in response to the question of whether the requirements and restrictions of data protection are seen as an obstacle to the introduction and development of Big Data. Quite the opposite: "Swisscom does not consider the requirements and restrictions to be a hindrance, but rather necessary and sensible. Data protection must not be softened in order to preserve the privacy and associated freedoms of citizens."
Christoph J. Walther
