From probabilistic to predictive analytics

Is that a crystal ball in your pocket or are you just crunching data?
BlabPredicts2_predictive analytics

This story is part of our Next Big Things series that appears in the September 2014 issue of strategy.

If it were up to Alex Glinka, he’d have the industry be honest and stop playing oracle. It’s not that the director and head of technology at FCB Toronto thinks brands can’t calculate that a person from Calgary will purchase more bug spray when the temperature rises by a degree, he would just rather call it a probability as opposed to a true prediction.

“We can create data models to extrapolate the probabilities of certain outcomes, and then apply a confidence level to whatever that hypothesis is, but we can’t be certain that this person will buy this widget because they bought this other widget,” he says.

However, we may be getting closer to oracle status with companies like Blab and its product BlabPredicts. Having recently soft-launched the product in Seattle earlier this year, the third-party service spends its time monitoring the social space. By using algorithms to anticipate the growth potential of conversations that will take place across six different social platforms and more than 50,000 news and blog sites over a 72-hour period, the company is able to give marketers the power to take action in real time. The most obvious outcome is that brands are able to alleviate consumer grievances and minimize negative press before they get out of hand.

There are other companies that help brands resolve customer issues before they implode online, such as New York’s Sprinklr, a social media management co that similarly analyzes conversations and finds trends for brands to jump on.

Mitch Joel, president of Twist Image, believes predictive analytics will become even more intelligent and eventually enable brands to nip an issue in the bud, before there’s even a reason for consumers to complain.

Only then will analytics go from being probabilistic, as Glinka describes, to truly predictive.

“For example, say I’m flying home and my connecting flight gets cancelled, the technology should eventually be able to not only notify me, but book me for another flight right away, and then provide customer service with a free spot in the lounge,” says Joel. “That’s an actual pre-emptive strike that prevents me from going online to complain.”

But before marketers can play in a predictive space, there needs to be an adoption of systems that effectively brings together multiple touchpoints, from customer service to online shopping, so that there is a single, unified view of a consumer, says Joe Dee, VP of product and technology strategy at Cossette.

“Someone is in the market for a phone, which a brand can see based on web analytics. That person calls in to be serviced for something else. You can then look at that as an opportunity to engage them on their desire to purchase a new phone,” he says, adding that Amazon and Apple are two companies that have the infrastructure to learn about their customers across multiple platforms and determine if and when they’re in the market for a particular product.

“The challenge is that these are digitally-born technology companies, so culturally it’s in their DNA, and they create products and services that inherently support that,” Dee adds. “Getting most brands in line with creating a holistic view is challenging because they’re traditionally siloed and may not know how to unify their customers’ data to give back relevant experiences.”

He points to the emergence of data management companies like U.S.-based Knotice, which was recently purchased by IgnitionOne, to help companies stitch together data across platforms and eliminate friction between touchpoints.

When Joel thinks of the role robust predictive analytics will play in marketing, he sees communications becoming efficient to a point where there are lower costs involved in acquiring customers, which means more returns for brands to focus on building and promoting their products.