Recently I put my house up for sale on the market. A few weeks later I started to get flyers and adverts in the mail for moving services. Interesting I thought. Someone, or some system, somewhere had anticipated that I might need some help shifting the clutter that I've gathered over the past 6 years.
The above is just one example of what I call "predictable" analytic marketing, driven by the crudest of "if-then-else" analytic algorithms – i.e. "if house up for sale, then send junk mail".
That's fair enough business logic I guess. Predictable marketing is driven by the most generalist (and basic) of analytics. It's also the reason why I get invites to $5,000 IT strategy courses at Cass Business School based on the IT in my job title. Not that my company would ever pony up to send me on one mind you!
Admittedly we're talking about crude decision making criteria here. But with the abundance of customer data and sophisticated predictive tools available to companies, surely there's a smarter way to target even the most "shot-gun" of all unsolicited junk mail.
However this form of predictable marketing can be persistent, and annoying. For the past five years I have been getting offers in the mail to sign up for United Airlines' VISA card. Again logical since I'm signed up as a United Mileage Plus member. Yet each time I've declined – i.e. I've promptly shredded the application to be recycled as….another application in my mailbox!
The analytics field is littered with metrics like customer propensity to buy or churn? But is there a metric to address non-response and, perhaps more importantly, what action to take if it persists? If so then why doesn't the United Airline automated system that churns out credit card applications by the thousands each day pick up on that fact with me – and save a tree or two in the process? Surely five years on non-response must give the system a hint that I'm not biting.
The point I'm trying to make here (apart from encouraging United and others to keep bombarding me with offers for goods and services that I don't want) is that applying analytic insights without any recourse to revision or feedback on the success (or failure) of its actionability is much like being stuck in an endless automated if-then-else loop. Which is pretty much what I find myself in with United.
The "if" and "then" parts seem to be well covered. But what about the "else"? In other words how do analytic systems cope with rejection? Humans tend not to cope well with that. But computers are more thick-skinned. In United's case it seems to be persistence – or marketing by attrition, in the vain hope that I'll someday bow into submission someday. But there's also plenty of scope to analyze the rejection or non-response "metadata" to try new or cleverer ways to bring me on board, rather than more of the same.
Perhaps I could elicit a response from my good friend and fellow blogger James Taylor at Fair Isaac on how rules and decision management technologies can manage post-analytic decision maintenance for smarter decision making. Or in other words how predictive analytic systems can be more self-learning based on their success or failure, and adjust decision strategies accordingly. Without that critical feedback loop it's all really just predicable analytic marketing.
Leave a comment