The amount of information transmitted and collected today is staggering. In 2012 alone, it’s estimated that approximately 2.8 zettabytes worth of data was created. For comparison, just one zettabyte is roughly equivalent to one billion terabytes.
And with all this information being bandied about, for most businesses, the natural question that comes to mind is: “How can we use it?” That’s because the best (and easiest) way to make financially successful products, solutions and services is by knowing exactly what your potential customers want. And data lets you do just that.
So how might predictive data analytics change the way the world and businesses operate in the years to come? We sat down with Dr. Eric Siegel, former Columbia University computer science professor and author of Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die, to discuss predictive analytics and how it’s changing the way we do business now—and in the future.
How close are we to accurately predicting human behaviour?
In general, human behavior cannot be predicted accurately. But predictive analytics need only predict better than guessing to achieve great value for targeting, marketing, and combating credit risk and fraud. In my book, I call this “The Prediction Effect.” Because a little prediction goes a long way, especially when it comes to your bottom line!
What do you think a normal day could look like 20 years from now in terms of how our computers and devices might work for us and the type of data they’ll be capturing?
Like weather forecasting (which is only accurate a few days out) I’ll hedge my bets by predicting only seven years from now.
At that time, in 2020, mobile connectivity combined with the more pervasive deployment of predictive analytics will render the value of this technology much more apparent to the end consumer. For instance, take your car.
Predictive analytics is primed to change how we interact with the world.
As you enter your vehicle, an anti-theft predictive model establishes your identity based on several biometric readings. Your car’s navigator pipes up and suggests alternative routing due to predicted traffic delays. As you’re driving, an en route drive-through restaurant is suggested by a recommendation system that knows its daily food preference predictions must be accurate or you will disable it. Your social techretary offers to read you select Facebook feeds and other relevant responses it predicts will be of greatest interest to you. And if you’re distracted too long by a personalized billboard, your seat can even vibrate if its internal sensors predict your attention has wavered.
What are some unusual and/or exciting ways that companies or industries are using predictive analytics to disrupt their market? How can a company gain an edge over its competition?
To secure a competitive edge, prediction is the holy grail of marketing, credit risk management and fraud detection. The millions of per-person actions and decisions taken by organizations across consumers are rendered more effective by prediction, and allows you to foresee which customers will respond, cancel, default on credit payments, or commit fraudulent transactions.
Beyond these three major business applications of predictive analytics, many enterprises find unique ways to apply prediction specific to their business.
For example, Google uses prediction not only to serve up more relevant search results, but to predict, on behalf of their advertisers, which new paid ads will be perceived as low quality by end users. Other the other hand, insurance companies expand standard actuarial methods with predictive models to predict insurance applicants’ risk of health problems and car crashes.
In terms of measurable metrics and impact, marketing is often considered the hardest to track. What kind of data should one look at in order to decipher’s a campaign’s true return on investment (ROI)?
Predictive analytics concepts were used by President Obama’s campaign team to garner more votes.
News flash: A high response rate is not the business objective here because it’s not a clear indicator of ROI. Rather than observing how many of those contacted were successfully acquired as customers, the effectiveness of marketing is reflected only by how many of those contacted and acquired wouldn’t have been acquired otherwise—i.e. the incremental impact of marketing.
Predictively optimizing for this objective is called uplift modeling, a topic I discuss in my book. In essence, uplift modeling is the prediction of persuasion in order to plan persuasion. The Obama campaign employed uplift modeling (they called it persuasion modeling) to gain more votes for their candidate in the 2012 presidential election.
With the explosion of the mobile market, as you look into the future of business, how important is predictive analytics to mobile in terms of increasing revenue and customer engagement?
Beyond mobile applications, such as those mentioned earlier in my car example, it’s a fact that already today virtually all major telecommunication companies (at least in North America) apply predictive analytics to target retention offers. They predict which customers are at risk of defection in order to triage the use of resources for retention outreach, and the offering of more hefty discount (or free devices) for retention purposes.
The goal for most mobile companies today is to increase their revenue. What are some ways to increase customer conversion using analytics?
The name of the game in customer acquisition is deploying marketing outreach more effectively and more efficiently. You can always spend more to increase marketing outreach. But the real question is, how can you get more for the marketing dollar already in your budget? The answer is focusing your marketing more precisely on those individuals most likely to respond. That is, those assigned a higher probability with a predictive model.