Advertisers take note: the world of big data is at war with itself and it may affect you. As a marketer, you may not want to think about this, because it seems very technical and off in the ether. But on the other hand, if you want to go programmatic, as nearly everyone has, or you want to target properly, you must familiarize yourself with what’s happening in the the relatively new field of data analytics. Analyzing your own data can show you how to target ad campaigns and make your media buys most appropriately, but joining that data to ad inventory data may become more expensive or difficult in the future.
Probably the most famous example of data used this way was the Target pregnancy campaign. Target used data mining techniques to predict the buying habits of clusters of customers. They were able to predict customers who were going through big life events through changes in buying habits. Predicting customers who are going through big life changes such as marriage and pregnancy is huge for retailers, because these consumers change their buying habits and usually buy more than at other times in their lives.
Target was able to identify roughly 25 products, including unscented lotion and vitamin supplements, that when analyzed together, could help the retailer arrive at a “pregnancy prediction” score. That score was so accurate that Target knew some women were pregnant before their own families did! Here’s how conflicting standards could have altered that highly successful campaign: the platform Target used to crunch its own data could have been incompatible with the platform used to serve ads in real time, affecting either the ability to target the right customers or the ability to buy media for the right price.
Just as back in the days of Betamax and VHS there’s a conflict of standards, and the best standard may not win.
In the Hadoop world, there are three main players: Cloudera, the startup backed by a massive $740 million investment from Intel that says it topped $100 million in revenue last year; Hortonworks, which went public late last year in a $100 million IPO; and MapR, which has raised $174 million in venture capital. The market potential is large because the technology can be used any time a company needs access to huge amounts of data quickly. Online advertisers, for example, need to decide which ad to serve to which viewer within a few milliseconds, and they often use Hadoop to help sift many terabytes, if not petabytes, of data in nearly real time. And demand is growing: Hortonworks had only $46 million in revenue in 2014, but that number nearly doubled from the previous year.
This war of standards can cost you both money and convenience, because two competing standards mean someone else may have to get in the middle to integrate them or make them compatible, and that business opportunity will take a bit of your ad buy dollar and probably a penny from the publisher as well.