Monday, 27. January 2014

Predictive modeling

Everybody is talking about big data but somehow still most companies don’t know how to use the amount of data properly and add real value.

Everybody is talking about big data but somehow still most companies don’t know how to use the amount of data properly and add real value. Data-warehouse solutions like Cognizant and Business Objects were promoting their services already more than a decade ago.  The possibilities of gathering data independent from the source and combine and analyze the data as you need and do relevant reports on the fly. Now with all companies going on the cloud the companies have less and less to worry about different databases and formats etc. Most information they are gathering is somewhere on the cloud and ready to be used. The amount of data collected is incredible. Everyday an additional 2.5 quintillion bytes of data is created. What can we do with this information? Recent news about the NSA gives you an idea how this vast amount of data could be used and possibly exploited. However the data could be also used much in favor of the clients if done properly. Complicated algorithms from Netflix, Amazon, Apple are already suggesting what you could be interested in based on your previous media consumption or your entered search terms. The worse the algorithm the more the client feels like somebody is simply trying to sell him something. The better the algorithm and the less intrusive the way the suggestion is presented the bigger the perceived value added by the client. Sometimes, I am simply puzzled and wonder how did Google, Apple, Spotify or Amazon know that I was looking for this information, movie, music track or book since a long time?


Of course you might call me a geek but sometimes I am simply trying out the suggestion from my apple tv how well the algorithm has predicted my movie preferences. The hardest movie to predict by algorithms is apparently Napolean Dynamite. You know the movie? This movie is either loved or hated and there is no between. It is extremely hard to predict who likes the movie and who doesn’t. The first time I saw the movie I was laughing a lot but frankly the second time I already felt bored. Anyway, the algorithms are getting better and better and the more you can rely on them the more relevant they will become for our future decisions. This is where predictive modeling comes in handy and will influence our everyday live more and more.


How far predictive modeling is can be observed for instance when a Target store manager knew before the father of the 16-year-old pregnant child’s about the pregnancy of the daughter. Impressive how war predictive modeling goes. The only thing that Target did was automatically analyzing the purchasing patterns of the teenager girl. Some people might also call it frightening. The father of the girl did not believe the store manager and told him that he was wrong. Well, he wasn’t. Some might call the power of predictive modeling scary. I guess therefore I still do not use loyalty cards because I think it nobodies business what I purchase. Nevertheless I see the potential if things are analyzed properly and the correct conclusions are drawn based on the data and if the result is presented to the client in an appropriate manner.


At the moment the majority of the people feel uncomfortable about the pre-filled search terms in Google, the suggestion from Google that according to your e-mail there should be an attachment attached to the mail or the advertising feeds in your Facebook newsfeed that still feel mostly like at random. However, the more data will be on the cloud and the more data we will share these algorithms will become more intelligent and hence value adding. Nevertheless, I think despite all the server power you put inside a human touch is still needed. Or would you blindly trust the weather forecast without even watching outside the window and make sure whether you need an umbrella or not? I think we are on the right track and that with the help of all the available we can actually help people to make better choices. All this available data can not only be used by modern marketers but by companies and help people and other companies to be more efficient and bring the right information and/or service at the right instant to the right consumer.  This is value-adding.