Table of contents for IBM Business Analytics Summit 2011
- Charting an analytics course to better outcomes with IBM
- IBM Analytics customer stories
- IBM and delivering differentiated client value
- Predictive Analytics spotlight from IBM
- Social Media Analytics with IBM
- Smarter Customer Analytics and Decision Management with IBM
- IBM thoughts on Predictive Analytics Futures
IBM sees Social Media Analytics as part of Customer Analytics rather than as a separate category – the reason for analyzing social media is to better understand customers and what they are thinking. It is not enough, says IBM, to understand and analyze social media. This understanding must be integrated with other kinds of analytics to drive better decisions. Social Media Analytics to IBM then involves the use of insight derived from social listening AND the use of predictive analytics embedded within business processes (using decision management one imagines as decisions are at the heart of your processes).
IBM sees a social media analytics curve. Companies first “listen” to social media then integrate with transactional and other data to move to predictions that can be integrated with customer profiles and thus drive appropriate actions/decisions that are integrated into business processes. Social media analytics can be used to measure the effectiveness of campaigns, integrate attitudinal data from surveys with social media, predict customer sentiment to see “soft” impact of offers made and more. Cognos Consumer Insight is IBM’s social media backbone.
Cognos Consumer Insight analyzes data from multiple sources and provides a number of capabilities such as standard reports on volume and evolving topics are included as well as sentiment and affinity analysis. It is an on-premise solution that uses a third party service to pull in data from all the various social media sources. IBM has built the product to be very robust and scalable and believes its research group has provided some strong sentiment analysis algorithms.
The interesting part of all this to me is the ability to bring social media data and analysis to bear on the construction of predictive analytic models and business rules in transactional applications. For instance, using social media analytics to understand that a particular offer seems to generate negative sentiment even though it also generates sales and feeding this back to reduce the “value” of this offer to my cross-sell decision service.
IBM showed how the insight from social media analytics can be integrated with SPSS. One of the key things in social media is the search for advocates. Typically 5% of those participating are “engaged authors” and regularly respond to the comments of others while 75% are casual, only posting once in a while and the remaining 20% just passively listen. Predicting who might become an advocate can be a powerful tool as it would help focus outreach and marketing on those who can be converted to active advocates. One customer for instance clustered online personas (bloggers) into various segments and then built a predictive model showing how likely a particular campaign would be to drive bloggers into more valuable segments. This model then drove campaign choices. Another example might be determining that discussions of coverage is a big predictor of churn and rebuilding a predictive model to include social media data so that this can be included to improve the lift of the model.
Cognos Consumer Insight outputs can be loaded directly into SPSS Modeler as data sources, allowing them to drive predictive modeling and allowing them also to be integrated with other data sources like survey results, sales data, CRM data etc.
IBM has also done some work on integrating this with Cognos 10, Unica and Coremetrics.