Home » Ethical Data Mining: A Vital Component of Modern Business

Ethical Data Mining: A Vital Component of Modern Business

Ethical Data Mining

The term Big Data is hyped quite a lot these days. But, even if it is overhyped, a bunch of ethical issues related to privacy, confidentiality, transparency, and identity rises up due to the Big Data revolution going on in the industry. Who owns all that data, being analyzed? Are there any limits to what kind of inferences that can be made, or what decisions can be taken against people involved in those inferences? Maybe these are the questions that boggle the mind when thought about these issues.

Before the Big Data revolution, the fields of marketing and advertising were guided by codes of conduct for many years. But, these codes of conduct apply to traditional businesses only. With the pace at which businesses have developed, these codes have become irrelevant.

In the last decade, the rapid emergence of Big Data, while clearly demonstrating obvious value for consumers and businesses alike, has incited concerns over the ethical use of consumer data in modern multi-channel marketing. In this high-speed chase of evolving environment, the absence of established ethical guidelines has caused anxiety among consumers and confusion amidst marketers — who are some of the most avid users of data and are also swift in the adoption of new data technologies.

Employing Sensitive Data for Marketing and Advertising

Previously, the marketing industry had simple terms for sensitive data. This data was about children, health, and finances. Today, every type of sensitive data is collected, such as location data and biometric data (e.g., facial recognition data), which can be very revealing about the activities and relationships of the consumers. Moreover, through sophisticated analytics, companies can take data that is not sensitive at all and predict, to a high degree of accuracy, very sensitive insights about individuals, such as the status of their pregnancy, what kinds of diseases they are likely to have or develop in the future, and what their financial situation is. This is, therefore, raising the paranoia among the customers, fearing it may get into the wrong hands.

Degrees of Sensitivity

The Direct Marketing Association (DMA), Digital Advertising Alliance (DAA), and Network Advertising Initiative (NAI) have set some guidelines about the usage of sensitive data. Use of third-party behavioral and multi-site data is to be limited to anonymous or opt-in consent based usage, according to the codes of conduct. Every code of conduct has certain restraints or prohibitions on the use of this sensitive data in conformance with concerned authorities.

The issues regarding the misuse of data tend to become more challenging when companies can predict sensitive data from non-sensitive data with sophisticated analytical processes. This practice provokes a whole new set of questions about predictive results. For example, the question arises whether to treat predictive data about sensitive medical conditions different from other forms of fact-based sensitive data? And should the individual have a say in the prediction?

Sometimes, prediction of the sensitivity of data when it is first created before all the various applicable uses can be understood, prove to be difficult. The full recognition of the resulting sensitivity might take years. Some data seem non-sensitive at first, but when collected in very large quantities over time, it might become predictive (e.g., NSA metadata, location data, biometrics, and other tracking data collected over time).

New Challenges Sticking Out Of The Hay

As the world of traditional direct marketing and advertising merges into multi-channel marketing and targeted advertising is driven by much richer data on individuals and devices, gaps in the ethical guidance codes have emerged.

There are three distinct areas where gaps in guidance exist in the U.S. and abroad:

  • Sharing data for marketing and advertising purposes (both personally identifiable information and anonymous information)
  • Employing sensitive data for marketing and advertising purposes
  • Applying robust analytics (e.g., Big Data analytics) for marketing and advertising purposes

Sophisticated analytics have led to superior data, powering innovation and significantly enhancing results.

The analytics discussed above approaches result in a prediction and, while they are usually quite accurate, there are always a few people for whom the prediction is wrong. Instead of relying on old methods, these predictions prove better than no data, for identifying audiences for marketing and advertising purposes. Pointing out that these kinds of predictions may not be as successful for non-marketing purposes is also important to sophisticate the overall process. Back in the day, these kinds of predictions were calculated over the certain period of time and stored on a file. More often today, they are calculated on the fly based on the most recent data, possibly data generated minutes or even seconds before the calculation.

Sometimes these analytical predictions are made by the company using them. In other instances, a third party company makes the analytical prediction and sells it to the company needing it. In other words, a social media site or a data broker might make a prediction with their third-party data about whether a consumer is in the market for a new Luxury car and sell that prediction to auto dealers with Luxury car inventories.

Consumer Concerns

There are certain concerns in the minds of consumers, about the safety of their personal data and the consequences of it falling into wrong hands, such as follows.

  • Consumer concerns vary widely from individual to individual, but generally, involve a lack of understanding of the analytical processes and how they are used. They can sometimes be viewed as a “black box” over which the consumer has no control, and are therefore very scary. The box shouldn’t be too clear about the algorithm or easily challenged if it is wrong. Consumers don’t know who is performing these analytics, what predictions they are making, or to what end they will be used. If analytics of non-sensitive data can create sensitive data, the question arises: Are proper precautions being taken with it? If sharing makes consumers feel uncomfortable, and then analytics can feel really creepy too. Consumers wish to know in which profiles they fall in, and some even worry that they are being profiled in such ways that might limit their opportunities or even discriminate against them.
  • In just the last few years, the proliferation of analytics and the number of companies who use them have grown exponentially. Sophisticated analytics have piloted to superior data, which helped power innovation and significantly got enhanced results. And while the people in the industry understand many of the extraordinary benefits that derive from Big Data analytics, consumers often feel that these practices have taken data about them to a new level of “out of control.”

Remedies

After considering the concerns of consumers over the five distinct areas where stronger ethical principles should be developed and introduced in marketing, the following broad recommendations are brought up.

  1. Maximize transparency and choice: Privacy policies of marketers must be clearer about the use and sharing of consumer data.
  2. Classify data and mitigate use risks: Marketing data should be classified to identify various types of risks, and appropriate mitigations for these risks should be put in place.
  3. Limit downstream risks: Data brokers should have contracts with all downstream users of the data to ensure it is always used appropriately.
  4. Help enforce ethical practices: Everyone in the marketing ecosystem should assist regulatory authorities in enforcing ethical practices by reporting bad factors in the industry to the appropriate enforcement bodies.
  5. Educate consumers about common marketing practices: The marketing industry should buoy and absorb in developing education for consumers about regular marketing practices.

In the more recent past, the DAA, comprising of the seven largest marketing and advertising associations, introduced standards consistent with NAI’s guidance for network advertisers, but which were designed to address the wider advertising ecosystem, needing the use of the “AdChoices Icon” in ads that were behaviorally targeted. This icon links to an industry website that lists all of the network advertisers, and offers consumers a simple, industry-wide, opt-out mechanism. Over one trillion times each month ads are served with the icon and 35 million unique visitors check out the site, with more than 4 million exercising their choice. According to DAA’s reports, 51% of consumers say they are more likely to click on relevant ads featuring the icon, and more than 73% are more comfortable knowing that the companies are abiding by the safeguards.

Establishing Ethical Guidelines for Modern

As the world ushers into the era of mobile and addressable TV, the DAA has once again updated its ethical standards, releasing mobile guidelines consistent with their online principles and launching an app, so that consumers can manage third-party tracking across apps. The self-regulatory and legislative tenets — notice, choice, and security — of the marketing industry, which was built on the sharing of data, have been consistent for decades. Each month over one trillion ads are served with this AdChoices Icon, which enables consumers to exercise control over behaviorally-targeted advertising. 51% of consumers say they are more likely to click on relevant ads featuring the icon thus serving the purpose of making consumers feel safe about the data sharing. Ultimately, these safeguards set in place are the ones making Ethical Data Mining comfortable to use in the modern businesses.