It’s no secret that data mining is the next frontier that will lead marketers to campaign gold.
When Orbitz discovered that Mac users spend, on average, $20 to $30 more for a hotel room than PC users, it began placing pricier rooms at the top of search results for customers on Apple products. Orbitz turned a seemingly innocuous point of data about its customers—which operating system they use—into a veritable treasure trove.
However, there’s a big difference between casting a wide net and a single fishing line when gathering information about your customers. While data mining is helpful for one-to-one customer relationship marketing and understanding individual customer intent based on Web activity, it doesn’t always scale well. When you bite off more than you can chew, it’s easy to become blinded by the details of your customers and lose sight of what truly matters.
Even industry leaders with top-notch personalization and data mining techniques can become too smart for their own good. Hoping to entice mothers-to-be to purchase products, Target devised an algorithm that assigned a “pregnancy prediction” score to customers based on past purchasing habits. The algorithm might work a little too well: It once homed in on a pregnant high school student and sent coupons for cribs and baby clothes to her parents’ home. Although the teenager hadn’t yet told her father about her pregnancy, the marketing materials sure did.
Amazon also has a world-class marketing program at the product level, but software can’t account for cultural and social overtones. For example, as Amazon’s sales of Confederate flags skyrocketed in June, curious people logged on to Amazon to watch sales climb. The e-commerce giant’s robust marketing program took note and began retargeting visitors, resulting in Confederate flag ads dogging people wherever they went on the Internet. Amazon even populated shoppers’ Facebook News Feeds with Confederate flag ads before removing the items from online shelves completely.
While data can hold the key to a profitable marketing campaign, falling blindly for the numbers can just as quickly land your brand in turmoil. To win with data mining, you have to be aware of the perils.
When Data Mining Becomes Dangerous
As these cautionary tales illustrate, high-caliber data mining technology won’t always lead you to the most appropriate customer data. Here are a few situations where data mining only depletes your time and resources:
• Collecting for the sake of collecting: Big data holds some fascinating trends. But it can get to a point where what you can find and learn trumps what you should find and learn. You need a clear case for creating actionable events from the data you collect for end users, whether that’s understanding customer behavior, learning how to better tailor your marketing, or seeking out the right opportunities for your business. Always remember that the end goal is to sell more of your product. Any data that doesn’t lead you to that goal is wasted time and money.
• Spinning your wheels: What if? What could we? What about? These are all telltale signs of a data quagmire. While these questions can help you get the ball rolling, don’t sit there and brainstorm the endless directions in which you could take your mining efforts. Don’t let data gathering become a blind hunt; realize when you have enough to know your customer.
• Analysis paralysis: If you find yourself pushing back deadlines or requesting additional bodies to solve overly complex problems, you might be in over your head. Analysis paralysis and scope creep are both real and present dangers when handling big data. Data mining can quickly become an overwhelming and budget-busting pursuit.
Keep Your Eye On The Prize
So how can you keep the big picture in mind while making personalized CRM scalable? With the help of big data trends, personalization scoring models, look-alike modeling, and a few best practices, you can narrow your scope, avoid analysis paralysis, and dig up only the customer data that matters. Here’s how:
1. Devise a plan from the start: Remember the scientific method? Use it to stay on track with your data digging. Before you begin, create an outline of your plan: What question do you want to answer? What data do you need in order to answer that? What do you believe the answer will be? Did the data prove or disprove your hypothesis? By aligning your efforts with these answers, you can zero in on the most telling information.
2. Learn to prioritize projects: You can’t approach data mining like a side project or one-off request. Let your marketing objectives guide you to allocating your data mining team’s time toward specific projects. Blindly digging through the numbers without a formal process wastes everyone’s time and only guarantees that nothing will get done quickly or properly. Define a process first to ensure that all employees’ priorities are aligned from the outset and that each person understands why his or her contributions matter.
3. Summon the right people to mine the data: I’ve encountered countless situations where somebody with the wrong skill set is tasked with data mining or customer analytics. In some instances, leaders will assign data-related tasks as a sort of “extracurricular” if someone shows even passing interest in customer data. While this might serve as a learning opportunity for some, it doesn’t put data mining in the most capable or dedicated hands.
Although data mining is enticing, don’t make the same mistake that Target and Amazon did and drown in your own data. By sticking to your initial road map, prioritizing high-value projects, and dedicating the right resources to the job, you can unearth the data gold that will inspire your next successful campaign.