What's Dragging Predictive Analytics?
So far, the story of Predictive Analytics has been one of unfulfilled promise. In this blog post and the next, I'll talk about the reasons for its slow takeoff and ways to accelerate adoption.
One of the things holding back predictive analytics is a lack of skilled resources. Working on a predictive analytics model is a specialised job that only statisticians can do, and they are hard to find. However, a lot of surrounding knowledge comes from the business domain, which can be leveraged by business users with the help of technical experts.
The second barrier is unavailability of good quality past data, essential fodder for any business intelligence solution. Where data is not readily accessible in the desired format, it may be necessary to extract it from the organisation's data warehouse - most have them these days - or in the worst case, pull it off their operational systems. Painful no doubt, but well worth the effort in the long term.
Today's businesses are information intensive and highly variable - this means that setting up an analytical model takes time and effort. Packaged third party offerings are emerging as an alternative to an in-house ground-up solution, and are worth considering for their expediency.
The other challenge to the adoption of predictive analytics is its slower payback. Predictive analytics delivers solid returns, but over a period of five years or longer and more so on big-ticket (read big investment) projects. This puts off many organisations lacking the necessary resources, risk appetite or patience. This is also the reason why pilot projects don't get the visibility they deserve - the bigger ones take too long to show results and the small ones don't show enough. One way to solve this problem is to involve business users right from the start, rather than leaving it all to a parallel team.
My point is that predictive analytics - like any technology - has its issues, but there are ways to help it along. I'm leaving that for my next post.
