We’ve all met a few: the master sales person. Armed with the same product and the same audience, these people outsell their peers by orders of magnitude. Clearly, personality and looks are a factor: they appear more trustworthy, more credible and forthright. But there’s something else. They’re also experts at reading a prospect’s signals – observing small details that provide clues not only about propensity to buy, but about what selling approach is most likely to persuade. For example, the sales master, observing a prospect’s style of dress, may adopt a hint of drawl in his pitch. He can see from the car seat she has kids, but he asks anyway. Seeing her check her watch, he cuts right to the bottom line. Sometimes he observes a signal that a prospect is likely to make a poor customer and quickly decides not waste time on them, even if he knows he can make a sale. And he’s always quick to pick the best targets out of a crowd and hone right in, using his secret scoring system.
This type of observation may seem like quintessentially human intuition. But the masters of digital marketing are using predictive analytics to replicate this type of insight at Internet scales. In the white-hot world of programmatic marketing (see Facebook Goes Programmatic), computer algorithms are scoring each available impression and interaction in real time using predictive analytics that evaluate thousands of factors, including whether the visitor has been on the advertiser’s website in the last 30 days, the weather, geography, time of day, device, competition, etc., in a closed-loop process that constantly searches for correlations between observable factors and desired results. Using predictive analytics, and a little help in the form of human creative messaging, computers are learning how to sell, and the results prove it.
The average digital marketer might consider a few factors – such as recent visits to the web site or demographic affinity with best customers or behavioral “in market” status – and manually set rules for what such impression are worth bidding for, when and where to send offers and what to show first to a visitor, making manual adjustments over time. Predictive analytics, if used at all, remains siloed off in a planning group that suggests new rules and trends to watch for. They may be using programmatic techniques, but they’re far behind the masters and losing ground every day. It’s especially hard for large, established organizations to connect predictive analytics to real-time operational systems that themselves are often siloed and protected. But, as Dana Anderson, SVP and CMO of Mondelez International said at Advertising Week, “if you’re not a start-up, you’re a turnaround.”
Finally, it’s not just organizational inertia that stands in the way of applying the power of predictive analytics to real-time digital marketing. We’re all aware that the image of marketing skating on the edge of influence engineering can seem unsettling, even dystopian – just as we’re uncomfortable around those master sales people, once we realize they’re craftily assessing our every quirk, it’s much scarier to imagine computers doing the same thing at scale. Such mastery treads a fine line between using observation to really offer the best value to customers who will derive long-term benefits from a personalized relationship with our brand, and using observation to simply maximize sales by taking advantage of the subtle but inherent predictability of human behavior. One can hope that the open forum of social media, by exposing manipulation for what it is and making us all smarter customers, will steer organizations toward the former approach. Some great sales people are also honest, I think.