"Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Ray Kingman, CEO at Semcasting.
I get it. Starting with our first-grade report card and continuing to the last time we pulled our credit report, we’ve feared someone else having control over what is being said about us – and who can hear what is being said.
Many think that any information gathered about us should always be private. And there are real threats to our privacy that we should be rightfully concerned about, including credit card number theft, surreptitious tracking of our locations orthe selling of personally identifiable information to the highest bidder. I start to worry, however, when you scratch the surface of hyped privacy intrusions only to discover they are largely scare tactics.
The World Privacy Forum, for example, just went public with its concern that “hundreds of secret consumer scores” are being used to predict consumer behavior. Financial, retail and other commercial enterprises are engaging in the unregulated use of predictive models to “index and score” consumers in order to make smarter decisions about how to engage with their customers and prospects. To add fuel to the fire, the FTC has chimed in offering to investigate these consumer scores, conflating them with credit scores and its ongoing concerns regarding the big data practices of data brokers.
The scores at issue use a statistically derived approach to project a person’s propensity to do something. It may be your likelihood to purchase a product, click on an ad, vote or pay your bills. Scores are nothing more than mathematically informed “guesses” suggesting how you, or people who may be similar to you in some way, may behave given a certain set of circumstances.
Unlike credit scores, which are based on financial modeling and can only be legally used under regulatory control to grant or withhold a purchase on credit, the consumer scores in question are nonbinding projections of behavior. They are not a fixed metric that can keep you from being approved for a car or a home loan. They are of no value in determining the interest rate you are offered or the size of the loan you can secure. These scores have one purpose: to statistically “guess” whether or not the product being offered is a good fit for you.
The assumption is that the better the fit, the more likely it is that you will buy.
How have predictive marketing models become conflated with credit reports and consumer privacy? This privacy discussion has covered a lot of ground, from the FTC and Do Not Track to the White House and the Consumer Privacy Bill of Rights and, most recently, to the report to the president on big data and privacy.
In the beginning it was about ad networks invading people’s privacy by tracking them online and recording what they viewed without permission. But after a flurry of calls from Silicon Valley, the Bill of Rights was watered down and the privacy issue seemed to simmer down.
The issue heated up when Edward Snowden and credit card breaches at big-box retailers hit the news. Senators Rockefeller (D-W.V.) and Markey (D-Mass.) then opened up a new front by taking aim at data brokers with the Data Broker Accountability Act. A 60 Minutes segment, “The Data Brokers: Selling your personal information,” followed. At this point, the public – and regulators – reached a rolling boil and were ready to go into full privacy-conflation mode, where mentions of data and the Internet created a privacy scare.
It is an instinctive reaction to be concerned about personal privacy. But when I hear about the WPF and FTC going after analytics, predictive modeling and the scoring of consumer propensities, I wonder if we’ve gone too far. If we are ready to equate normal marketing best practices and simple audience segmentation with credit card theft and the unauthorized tracking of online customers, what’s next?
If marketers were prohibited from using public data or any form of analytics or scoring to identify their best customers, what would happen to retailers who couldn’t connect products to customers most likely to buy them? Or auto dealers that couldn’t use consumer data and analytics to separate Mercedes buyers from Hyundai buyers? How about political candidates not allowed to use data and modeling to measure the intensity of issues that matter to their constituents, or to improve get-out-the-vote efforts or target high-value donors?
There are scores for every consumer in relation to every make and model car, every category of book and movie and all the various smart phones, appliances, window replacements, lawn mowers, weed whackers and so on. Scores are how marketing works. Without scoring, there would be no way to do intelligent segmentation. Without segmentation, CRM wouldn’t even be an acronym, and marketing automation would be nothing more than a box of pre-addressed FedEx slips waiting for a customer list.
Finally, there is what marketers like to call the long tail. Without being able to predict our preferences or score our projected commercial intent, the efficient engine that drives a big part of the economy today would grind to a halt. The long-tail model that allows us to receive targeted offers and transact efficiently with favorite retailers, the best shoe stores or our iTunes account would disappear. Instead, every retailer would have to market to every customer in the same way, kind of like the printed version of the 1960s Montgomery Ward catalog.
We do need to acknowledge that there may be instances where the use of public data and the application of scoring create the potential for damages to the consumer. Such situations should be subject to regulatory guidelines, just as with credit scores.
In these cases, regulators and marketers alike should ensure that the damages from data collection, modeling and scoring justify any action taken. This is because the loss of public data, first-party data collection and analytics would lead to dramatically less efficient and, for the most part, unprofitable consumer marketing.
Usable consumer scores based on predictive models that combine retailers’ data with publically available information have helped to create a more efficient, near-frictionless ecommerce model. This allows buyers and sellers to more seamlessly connect supply and demand, resulting in a more consumer-friendly marketplace. Whether you are a retailer, consumer or the FTC, that’s something we should support without further conflation.