Like many technical topics in the “big data” realm, marketing analytics is very exciting for insiders, but for most people it probably sounds like just another over-hyped, invasive way to sell a little more mouthwash. But I’d like to argue it’s actually one of the most under-hyped topics around. Its relative, privacy, certainly gets plenty of attention, but it’s predictive analytics themselves that I believe are right at the fulcrum point of utopian and dystopian visions of the future.
Technologies often find themselves in the balance: think of nuclear power, or genetic engineering. But while the effects – both good and bad – of most revolutionary technologies are generally tangible, the long-term impacts of predictive marketing analytics on both sides of the balance sheet are poorly understood and ironically unpredictable.
On the anxiety side, I’m not just talking about revelations that your smart TV might be spying on you or other scary data collection scenarios described a new report from the Federal Trade Commission which is trying to advance the long-sought cause of more consumer control over privacy and data collection by connected devices. There are many important issues to debate here, but even under optimistic forecasts of what government, business, and consumers might achieve through collective wisdom and self-restraint I think it’s safe to assume that exponentially more data will continue to flow into marketing systems (even if consumers do gain more control over its flow), and marketers will continue to advance the sophistication of their analytic capabilities.
The oft-cited benefit of this to consumers and society is the end of irrelevant and unwanted marketing. Critics who claim that more data collection will lead to more invasive marketing often seem to overlook this point. When our marketing analytics get good enough to predict when we’re most likely to respond positively to an offer or branded message, and our digital agents get smart enough to filter out the remainder, the economic benefit of blanketing our screens and mailboxes with annoying messages on the basis that even an infinitesimal response rate is worth it will fade away. David Ogilvy’s famous observation that “you cannot bore people into buying your product, you can only interest them in buying it” will finally be heeded and enforced by the marketplace.
Armed with smarter machines and better privacy controls, there’s much to be gained, both economically and socially, from using analytics to establish better trust between brands and individuals based on transparency, choice, and good customer experience. So, if we can master the privacy challenges, what’s to be lost?
Consider that marketing analytics is largely the science of scoring. There are many kinds of scores – credit scores, lead scores, affinity scores, and so forth, and there are dozens of algorithms behind these scores. To make marketing relationships scale requires thousands of split-second decisions, which must be automated, and automated decisions require a number.
Last year, the World Privacy Forum coined the term “consumer scoring” in a report called “The Scoring of America: How Secret Consumer Scores Threaten Your Privacy and Your Future.” Despite its undisguised bias, the report concedes that consumer scoring is not “inherently evil” and that, “when properly used, consumer scoring offers benefits to users of the scores and, in some cases, to consumers as well.” It then makes the case that scores should not be kept secret and that consumers should be given access to the data that makes up their own scores so that they can refute them and correct them. This sounds good, but there are a number of problems with it. Rachel Thomas, VP of Government Affairs for the Direct Marketing Association, described one: the fluidity of scores. “It doesn’t make any sense to call predictive analytics a ‘score.’ In marketing, the interests of consumers are constantly changing, so there’s no such thing as a static score,” (quoted in AdWeek, April 3, 2014). I’d add that, if the algorithms of advanced scoring were fully revealed to consumers as recommended by the report (“All factors used in a consumer score must also be public, along with the nature and source of all information used in the score”) the resulting thicket of text would make current T&Cs look like Dr. Seuss.
Aside from dangers associated with mistakes and abuse, the real problem with pervasive score-driven marketing is its systemic effect on those whom our algorithms “low-scorers.” Regulations may compensate for the incentives banks or insurance companies have to openly exclude high risk individuals, but what happens when millions of small automatic decisions about whom to offer what are made opaquely in the cloud? A common prediction is a widening gap between “desirable” and “undesirable” segments in every marketplace.
Perhaps there’s a future where ethics and economics naturally converge as smart machines reveal these effects and reward ethical practices, reflecting aggregated users’ social values. In any case, it’s becoming increasingly important to apply ethical considerations to our analytics regimes, if only to assure compliance with applicable laws and policies. This is a complex topic that many marketing data brokers such as Acxiom are taking on, but it’s safe to say it will require a good deal of effort to get it right.
Back at Gartner, we’re looking at what’s happening in marketing analytics and how marketers are using them today. Because, whatever you think about ethical implications, if your marketing isn’t advancing its analytics practice, your business could be in trouble.