Marketers who use the scientific method to make the most of “big data” or regular data (not as big, limited variety, and less velocity) run the risk of creating a hypothesis so myopic it leaves little room to reveal trends or other insights. A good hypothesis is built on previous research and theory, yet answers something new.
Assuming you’ve tested a non-myopic hypothesis, and reviewed and analyzed the results, what’s the next logical step? This is where more mistakes can be made.
In 2012, Orbitz found people who use Apple’s Mac computers spend as much as 30% more a night on hotels, so the online travel agency showed them different, and sometimes costlier, travel options than Windows visitors see.
Orbitz’s chief executive Barney Harford recruited a team of statisticians with backgrounds from eBay and Google for his data mining team. Those statisticians found hard facts such as nearly half of bookings on Orbitz for the high-end PUBLIC Hotel Chicago came from Mac visitors, so they included the computer’s operating system into Orbitz’s search algorithm.
Expedia, Priceline, and Travelocity do not include the person’s computer operating system when suggesting hotels.
Consumers using an iPad or iPhone to shop online may also see higher prices because retail websites like Rue La La know that iPads and iPhone account for 75% of their mobile sales.
Is the person’s computer operating system really why those individuals spent 30% more on Orbitz.com? Perhaps there were other factors, processes, or pathways that would explain why those people paid 30% more. They just didn’t see it.