Predictive Intelligence: Mad Data Science in the Inbox >

surprised dude_resizeImagine a world in which we can harness behavioral data—everything from what size you click on when online shopping to what types of blog posts you compulsively click—to serve up a different, completely personalized email for each and every customer.

Reality: This can happen today, for your email program, though few are putting this slick data science to work.

We got a full download on the Predictive Intelligence add-on in Salesforce’s Marketing Cloud recently, and were pretty darn impressed with the feature, which allows users to personalize email programs at scale. Basically, this tool builds a profile for each customer in real time, then, with fancy algorithms, predicts what their next move will be.

We took a look at our own inboxes to see which brands utilize such “intelligence” and whether it can make a difference in consumer behavior, admittedly with a sample size of 1. After a little bit of “research,” we found intelligent optimization is a rare feature in the inbox. Here’s an example of Predictive Intelligence done well and… not so well.

Etsy_Gmail copy

Predictive Rockstar: Etsy

Etsy is extremely skilled at surfacing cool, crafty products that the customer is interested in through their app, website and, most importantly, in the inbox. In an email we recently received, items of potential interest were surfaced without requiring us to answer any questions. “Inspired” suggestions begin in the inbox and continue into the app or website, ultimately providing a customer with lots of suggestions until something catches the eye. A hassle-free, personalized experience – it CAN happen.

Screen Shot 2016-03-09 at 12.49.14 PMNot-So-Predictive Smarts

Almost 3 years ago, a BrightWaver bought tickets to see Jimmy Buffet from a popular ticket reseller. For months, she received emails recommending some pretty decent musicians—none of them really Buffet-esque. The predictive model hardly made a compelling case our staffer to continue engaging (engagement here could mean opening an email, clicking on hyperlinked content and so on). In fact, she started ignoring the content of these emails and used the inbox notification as a reminder to check the site for tickets to that next epic concert. (And that’s the best case scenario. Most would simply unsubscribe if the content isn’t relevant.) Getting predictive intelligence wrong can be a poor reflection on the brand and show the consumer that you totally don’t get them—the exact opposite of what you set out to do.

In Conclusion: Predict Carefully & Often

You are taking a little risk when you surface content and say that it is “personalized.” People are complicated. This year I’m craving a new pair of overalls for summer as a reminder to my middle school years. But next year I may be ready to move into my 30s leaving all childhood memories behind. People want their email to reflect them as they are today, at this second. Luckily for us (and other email marketing pros), predictive intelligence can easily be altered in real time, based on changes in personality and preferences, promising intimate and advanced knowledge into subscribers without the hassle.


Leave a Reply
(will not be published)