Putting Data to Use
Would you want to hear about how many variables were considered and what type of modeling techniques were employed for everyday weather forecasting? What we want know is a nicely packaged simple answer to decide what to wear that day and whether we should carry an umbrella or not.
I am writing this article as I am flying (again), so let’s start with an example of an airline business. Let’s assume that there are a couple of empty seats in the business class cabin. Of those who are sitting in the economy section, who should get those coveted upgrades? Based on what? Ticket class or the price they paid for this flight? Loyalty measured expressed in mileage? Lifetime value in dollars? Past 12-month spending total? What we do know for sure is that the ground crew who would hand out that ticket with a one-digit seat number does not have to consider hundreds of variables in the data dictionary. Details of ticket class would be more than a full page. It is the data manipulator’s job to make it simple, properly reflecting business policies. Just call out the name based on score ranking and hand out the ticket with a smile.
How about for personalization effort online? Decisions should be made in less than one second to show proper content through the right screen. Human beings are not making that type of decision for sure, but even machines require neatly arranged data. More advanced retailers employ personas for such efforts, and personas are nothing but statistically summarized information coming from hundreds or even thousands of data points for the target. I have provided plenty of examples in this series already (refer to “Personalization Is About the Person,” “No One Is One-Dimensional” and “Segments vs. Personas”).
Why should it only be about online marketing? Why not offline? Imagine a restaurant chain that serves quality steak in a reasonable price range, equipped with more than decent wine list. Some chains like this maintain order-level data at a customer level. Such data are seemingly hard to crack, but let’s start with a personal spending pie chart for every customer, and create a dollar ratio among main dish, side dish, pre-dinner drink, wine and dessert. Added to typical transaction data including location, amount, payment type, reservation method and time/date of visit, we can use this information to start building all kinds of models and personas, such as:
- High-value customer (with high lifetime value potential and/or high frequency visit)
- Client entertainment (with irregular intervals in varying locations with higher-than-average spending on drinks)
- Wine enthusiast (based on percentage of total spending on wine, and amount spent per bottle)
- Family event (with regular intervals, such as birthdays and anniversaries, in a set location)
- Family with young children
- Small portions / Special diet
- , etc…
All of these are simplified answers coming from hundreds of data points per customer, which can easily be shared with the restaurant crew via handheld devices in forms of probability or scores. So, if they “know” that the customer who made a reservation at 7 p.m. is a high-power salesperson with an ample appetite for fine wine, the manager in charge had better greet him with a sommelier standing next to him this evening. And he does not have to have a degree in statistics, for he would just be a consumer of information. Just greet him by name and show some appreciation, with a smile.
So, still worried that data are being piled up, but they are not being used properly? Turn them into tidbits in fun-size packages. Anyone can understand what “70% chance of showers this afternoon” means. Answers like “80% chance of being a high-value customer” is no different.
Data are not used because they are hard to digest. It is the job of data players to make them digestible. I’ve seen users struggling with data dictionaries with hundreds of pages to make simple queries. Now, whose fault is that?
Stephen H. Yu is a world-class database marketer. He has a proven track record in comprehensive strategic planning and tactical execution, effectively bridging the gap between the marketing and technology world with a balanced view obtained from more than 30 years of experience in best practices of database marketing. Currently, Yu is president and chief consultant at Willow Data Strategy. Previously, he was the head of analytics and insights at eClerx, and VP, Data Strategy & Analytics at Infogroup. Prior to that, Yu was the founding CTO of I-Behavior Inc., which pioneered the use of SKU-level behavioral data. “As a long-time data player with plenty of battle experiences, I would like to share my thoughts and knowledge that I obtained from being a bridge person between the marketing world and the technology world. In the end, data and analytics are just tools for decision-makers; let’s think about what we should be (or shouldn’t be) doing with them first. And the tools must be wielded properly to meet the goals, so let me share some useful tricks in database design, data refinement process and analytics.” Reach him at email@example.com.