Predictive Analytics service by Amazon: reality or hype?

Crystal BallOne of my favorite quotes about life comes from Yogi Berra who said, “it’s tough to make predictions, especially about the future.” Yogi is absolutely right — it woud be nice to predict the future by simply analyzing our past behavior. If life was that straightforward, then every armchair quarterback of an investor would make millions off the stock market by simply looking at past stock performance. In reality, life is just more complicated and far less predictable.

I was excited to read an article in the New York Times by Quentin Hardy* about the potential analytics as a service offering from Amazon. While Mr. Hardy was somewhat unclear regarding what type of analytics service Amazon may provide, the article seemed to indicate that they are interested in developing a predictive analytics service. Predictive analytics is incredibly valuable to marketers as it allows brands to analyze volumes data, such as purchase history or social analytics chatter, to determine when will segments of consumers are most likely respond to a marketing tactic or to complete a purchase.

Mr. Hardy’s idea of an Amazon predictive analytics service is within the realm of possibilities. A prediction engine requires copious amounts of data to learn from and Amazon can easily use their data storage infrastructure (aka Amazon Simple Storage Service) to handle this responsibility. Amazon has already proven that they can build a service that appeals to mass markets: in early 2011 Amazon released Simple Email Service, an email delivery engine. While the service is still in beta, Amazon recently announced incremental improvements to the service, which indicates that they are committed to keeping such a service going. The combination of this and the fact that Amazon regularly releases new offerings on AWS solidifies the notion that a prediction analytics service may be in the cards for 2012. One thing to note is that if such a service does come to fruition, it will instantly face competition from two worthy opponents: Google, a cloud veteran that has been pushing their developer-friendly Prediction API since 2010 and Zillabyte, a stealthy startup with a user-friendly prediction service, is due out later this year.

Since there are no details on the service, it is premature to guess how it may impact the predictive analytics industry. While there’s certainly lots of interest, we simply have to wait and see.

*PS I’ve actually met Quentin at the Lithium UP conference last year. I know that he knows people who know things so I’ll give this rumor a 70% chance for happening this year.