Why You Need to Curb Your Obsession With Big Data

Blind faith in big data is robbing us of our humanity, and if we’re not careful, our obsession will ruin us.

By all accounts, 2015 was a big year for big data. Fields as diverse as medicine, law and even fashion have gotten the big data treatment of late. This is good news; a data-driven world is a smarter place. But that revolution comes at a cost, one that's best illustrated by advertising and media, both industries that pioneered big data years ago and spent 2015 grappling with an existential struggle between creativity and technology.

As it turns out, the choice between creative and tech, between what's subjective and what's quantifiable, between humanity and computers, is a false one. Like all revolutions, the digital one bends toward extremism. The old way was guesswork, the new science of big data is perfect beyond our wildest dreams. Extremes are neither healthy nor sustainable. More importantly, adopting such extreme positions misses the larger point — big data is the most powerful tool we have, but like all the tools that came before it, it's not a panacea.

Big Data Binging and Retail Stalkers
Search online for a pair of shoes and before you can add them to your shopping cart, you’ll want to pick up a restraining order just to keep the banner ads at bay. Even consumers who may not be so tech savvy are well aware of the obsession with data-driven targeting. In fact, consumers have known for years that the often clumsy application of big data is more akin to stalking than advertising, even if they aren’t familiar with lingo like retargeting, programmatic or cookies. This is the logical result of a big data binge, but it’s only getting more extreme. Collecting more data and baldly adding it to a marketing algorithm exacerbates the problem; some consumers find it creepy or invasive, others install ad blockers or learn to ignore solicitous sellers.

Beyond the detriment to a retailer’s relationship with its customers, the obsessive big data binge is troubling because it speaks to an approach that favors quantity over quality. Retailers know they can use programmatic to buy an audience on the cheap. In fact, it’s so cheap that many advertisers have overlooked the ad fraud and viewability issues that plague programmatic. Just because you can buy audience cheap doesn’t mean you should. And if you do, it doesn’t mean you continue as the impact gets diluted. Furthermore, just because you’re running your campaigns with big data doesn’t mean you’re driving at real business goals. Put simply: Retailers win by using big data to maximize the quality of their engagements, not by scaling junk and fraudulent engagements at the lowest possible CPM.

How Should Retailers Apply Big Data in 2016?
Rather than obsessing over how to collect more data, retailers should instead focus on activating the data they have. If the data a retailer collects can’t be activated to achieve a marketing plan or business goal, there’s little value in collecting it. In fact, collecting data that cannot be activated is detrimental because it distracts retailers from achieving concrete business goals.

Up until this point, marketers have largely used big data to isolate, analyze and influence sequential events. An internet search or site visit leads to a series of targeted display ads. Opening an email triggers an avalanche of more emails with better offers. For every consumer action, the programmatic application of big data offers a corresponding counterpunch. Effective retailers have found value in this approach, but only to a point because most consumers don’t follow the choreography of the media buyer’s playbook.

Instead of using big data to force an artificial narrative on the consumer, retailers would be better served by using that quantitative firepower to uncover connections that already exist. Rather than imposing the artificial construct of a buyer’s journey onto an equally artificial funnel and segmenting that world into neat channels, retailers should embrace the complexity of reality. Moreover, retailers should use big data to uncover the connections that enhance a more nuanced understanding of reality.

Do consumers really buy shoes or a new couch because they’ve been targeted with all the precision big data can muster? They don’t, although if we look hard enough for a cause-and-effect relationship and we have enough data, we can probably convince ourselves that the 16th banner ad (out of 20!) really did drive the sale. More realistically, retailers need to think of their marketing as a complete breakfast. They need to serve their customer on their terms, and they need to be able to engage in ways that are appropriate and, ultimately, more personal.

A retailer might serve that ad for shoes based on an internet search, but wouldn’t it be better off using big data to decipher the lifestyle/shopping pattern that corresponds to the prediction of a customer’s regular and ongoing need to buy new shoes? Similarly, a furniture store could respond to a website visit with a mail offer for a new couch, but if the mailer arrives on a Monday, what’s the point?

Here are a few insights retailers can apply to make big data manageable:

• Apply some human logic. Data by itself isn't logical, it’s just data. For example, data may trigger an event to happen — e.g., sending mail or email — based on consumer behavior, but if the marketer knows that customers respond better to a Thursday mail piece for furniture than a Monday email, then the ship date for mail should be timed accordingly.
• Develop theories and test them. Most marketers using big data search for trends or create systems to act. However, the data scientist employed by marketers should be tasked with hypothesis-driven testing, like "how many direct contacts does it take to trigger unsubscribe activity," and then use data to scale the findings and inform strategy.
• Establish a single currency for comparison. Data can be used to deliver insight into media, branding, performance marketing and many other things. Establishing a single touchstone (e.g., return on ad spend) will give all the data experiments a number to tie back to so marketers can evaluate the entire portfolio.

Lewis Gersh is the CEO of PebblePost, a programmatic direct mail company.