We have announced the full results of our Predictive Analytics in the Cloud survey. The results are available as a white paper and as a recorded webinar – go to smartdatacollective.com/predictive-analytics-cloud to register for all the deliverables. There were some interesting results and I thought I would share a few.
- The core focus for predictive analytics, and for predictive analytics in the cloud, is improved targeting and development of customers. It dominated the top outcomes from predictive analytics as well as the top areas of focus in both predictive analytic sand cloud.
- All five of the scenarios – cloud-based predictive analytic solutions (decisions as a service), cloud-based deployment of predictive analytics into SaaS applications, cloud based deployment of predictive analytics to on-premise applications, using cloud-based data in modeling and pushing modeling to the cloud – were seen as powerful with no obvious winner. None of them are that widely adopted yet but, as you would probably expect, pre-packaged analytic applications did best. The runner up was the use of cloud to embed predictive analytics into on premise applications – an interesting result that shows the importance of deploying predictive analytics not just building the models.
- Decision Management was clearly an important element for successful analytic adopters. We asked companies how they used predictive analytics and overall people were split between predictive analytics providing occasional insight and predictive analytics being tightly integrated in operational systems (the basis of Decision Management). But when you focus in on those who have already seen significant positive results from predictive analytics the percentage tightly integrating predictive analytics into operations rose while occasional use dropped. Among those transformed by predictive analytics a whopping 2/3 tightly integrate their predictive analytics with day to day operations! The power of decision management.
- These more successful companies also valued different types of data for building models. Near real-time and real-time data were seen as more important by the respondents overall but among those with more experience both batch and static data scored much higher – experience clearly shows that less volatile data can be valuable too.
- Finally a couple of surprising negative results. I really thought that more experience with predictive analytics would make people more tolerant of “black box” models but in fact the percentage who really wanted transparency in their models started high (well over half) and climbed to 80% among those with the most positive results so far.
- Even success does not make people comfortable with black box models it seems. On the cloud front I really thought that transaction based pricing – pay as you go – would be a big driver but it did poorly across the board. Reducing the demands on IT and empowering the business were what people were looking for from cloud. I think transaction pricing has a lot to offer folks with decisions as a service cloud-based solutions in particular but it’s not apparent that the survey takers agree with me.
Register at SmartData Collective for more. Thanks to Clario Analytics, FICO, Opera Solutions, Predixion Software, SAS, Teradata and Toovio for sponsoring.
