SAP recently acquired KXEN, a leading provider of predictive analytics technology, to complement SAP’s existing analytic products. I have previously written about both companies’ products and this week got a chance to chat with John Ball, CEO of KXEN, as well as a number of executives on the SAP side. This is an important acquisition so I thought I would take this opportunity to share my thoughts on it.
Prior to the KXEN acquisition, SAP was already making some progress into the predictive analytics space. It was expanding beyond its established BusinessObjects BI and new Lumira visualization products with SAP Predictive Analysis. SAP Predictive Analysis has some nice features and is well integrated with SAP’s powerful HANA in-memory database engine, it’s clearly a new player in the predictive analytics workbench arena that is largely targeted to data scientists The acquisition of KXEN changes this dynamic in several critical ways.
KXEN is aimed squarely at business analysts not data scientists or data miners. KXEN has a lot to offer those analytic professionals, especially in terms of creating and managing large numbers of models and doing variable selection, but its’ heart has always been on bringing predictive analytics to a broader audience. This acquisition will, I think, bring SAP round to this way of thinking – indeed it seems likely to have been a key driver for the acquisition. I would expect to see a greater focus on making powerful predictive analytics accessible to less technical users rather than on simply focusing on the existing data miner community. This will create an opportunity among their BusinessObjects and Lumria installed base as well as more broadly (of which more below).
It will be interesting to see how this plays out with respect to R. As surveys like the Rexer Analytics survey show, R is exploding in popularity with data miners and data scientists. SAP has done a fair amount of work to integrate R into its products. But KXEN takes a very different approach to modeling than the R community. KXEN focuses on encapsulating and removing complexity from analytic model development – on making core techniques work really well. The R community in contrast often seems focused on how many new techniques can be developed and used. How SAP resolves this tension, how it continues to support R without losing KXEN’s focus on business users and reducing complexity, will be critical to a successful integration of KXEN into the portfolio.
SAP is clearly interested in cloud – they are a sponsor of our Predictive Analytics in the Cloud research this year for instance and they have SAP HANA cloud solutions. KXEN is also making a big push into cloud, especially with its products that integrate with Salesforce.com. These are what I call “Decisions as a Service” products, decision-making solutions that run in the cloud and embed predictive analytics as part of the solution. Using the KXEN engine, these products build and update analytic models automatically and handle analytic best practices like A/B testing and holdout groups. SAP also has some solutions in this space, such as SAP Real-Time Offer Management. I fully expect this focus on the cloud, and on analytic applications in the cloud, to continue. What will be interesting is to see if the focus on supporting cloud platforms like Salesforce.com continues and what the balance is between moving analytic modeling itself to the cloud and delivering cloud solutions that include automated analytic models.
Operationalizing Analytics and Decision Management
Our primary focus here at Decision Management Solutions is on operationalizing analytics to effectively automate and improve operational decisions. Predictive analytics is a key part of a Decision Management solution stack but not the only part: Because policies, regulations, best practices and expert knowledge all play a part in decision making too, it is critical to bring business rules management systems to bear on the problem in parallel.
One of the most interesting things about SAP’s acquisition of KXEN is that it brings the KXEN product lines into a company with a powerful business rules platform. SAP’s core enterprise applications are increasingly integrated with its ABAP business rules management system (SAP NetWeaver Decision Service Management). Once this environment supports PMML, something I expect reasonably soon, analytic models developed using KXEN or SAP Predictive Analysis can be deployed to more or less any ABAP system no matter how old. This will let SAP customers drive the most advanced analytic decision-making into even their oldest enterprise applications. If SAP can leverage the ease of use of both KXEN and their business rules technology they should be able to give their massive installed base a huge boost, making predictive analytics much more accessible there by dramatically increasing the pool of applications really taking advantage of predictive analytics.
It’s going to be interesting to see how SAP pulls all this together. They have a great stack now with all the right pieces. As always after an acquisition they will need to integrate the technology they now have. Even more importantly they need to integrate their messaging and focus on educating their core application customers to make a leap forward to predictive analytics and decision management.
If you are interested in the various products mentioned, here’s a set of links to previous product reviews on the blog: