By Cindi Howson, BI Scorecard
I wrote last week that one of the big trends in big data was a resurgence in the use of search and natural language processing for making BI as easy as Google. Last week at IBM's annual Information on Demand conference, IBM announced project Neo, heating up competition in BI Search and cloud BI.
Neo starts with a simple search box in which users can ask a question. In this screenshot, "what is the relationship between Budget, Gross Domestic Sales, by Story type. Neo will present a list of possible data sources that can answer the question. For now, these data sources are restricted to data sets loaded to the cloud, in a DB2 columnar data store. IBM concedes that for the product to be fully embraced, Neo will need to support on-premise data sources as well, and has said that is part of the product roadmap.
Once the user selects the optimum data source, Neo generates an interactive visualization. For example, a user can refine the question by change "budget" to "units sold," for example. The visualization can also be changed to display as a trend rather than a bubble chart. In addition to the visualization, Neo generates a number of infographics (shown along the top)based on statistical algorithms that might be relevant. So even though the user didn't ask about seasonality, the data is showing there is a pattern with Fall being the best season for sales (early holiday shoppers, perhaps?).
Neo brings together a lot of intellectual property that IBM has acquired in recent years. The visualizations are powered by RAVE (Rapidly Adaptive Visualization Engine) technology and skills acquired through SPSS. The infographics are based on some of the capabilities in Analytic Catalyst, a module released in June of this year that makes advanced analytics easy for a casual business user. The natural language processing leverages Vivisimo, later rebranded Infosphere Data Explorer.
Not surprisingly, the initial demos of Neo are impressive. It's easy, visual, and powerful, for the most casual of decision makers. It could do for data what Google has done for the Internet. Today, the industry average for BI adoption is at 24% of employees, and ease of use is an oft-cited barrier to broader BI (take this year's survey to rate your BI adoption and ease of use).
What wasn't shown, though, is how the data sources get indexed and loaded to the cloud. IBM Cognos has previously tried to leverage the simplicity of search with its Go! Search interface launched in 2006. With that tool, content in PowerCubes and reports had to be indexed on a periodic schedule. Search was limited to key words, and the interface was existing reports, not nearly as visual as Neo. So Go! Search had a degree of complexity to implement, was less intuitive, appealing, and smart. Perhaps these are all reasons why it wasn't widely adopted? Just how well Neo overcomes the past limitations of Go! Search will only be known once the beta launches in January.
IBM also announced another new product, IBM Concert, that brings collaboration, workflow, and mobile together in a SaaS solution. This product is expected to be available in December. I see collaboration as still an emerging capability that customers are trying to figure out. Recognizing the influence that vendors like Facebook and Twitter have had on consumers, social and collaboration capabilities have began appearing in BI tools and enterprise apps a few years ago. SAP first launched Streamwork, then later acquired Success Factor's Jam, while Microsoft acquired Yammer, and TIBCO launched Tibbr. IBM has had strong collaboration capabilities in its Lotus Connections product, whose technology was first integrated with IBM Cognos in its version 10 release back in 2010. But I haven't found a single customer using those capabilities. Again, it's not clear if that's because it was poorly marketed and BI teams have other priorities, or if it reflects a larger, industry-wide problem. Collaboration around data today is usually offline from the data, whether via email or in meeting rooms or conference calls. So first, capturing comments in collaboration software is a change in the current work flow. Second, making comments publicly requires an analytic culture in which it's safe to voice opinions, dissentions, and to ask tough questions. Just imagine, if 30 years ago, the engineer who was worried about the Space Shuttle Challenger posted a comment in the data analysis of O-ring tests to the effect of, "the data shows we shouldn't launch. Too cold out." In those days, the engineer could barely voice a concern in whispers and only to his direct supervisor. (If you're as fascinated about that story, culture, and decision making catastrophe as I am, I'll be watching the Science Channel's documentary this Saturday). How far have we come since then? Is social something reserved more for personal and public opinion, or is it something that the industry is ready to embrace in BI?
I look forward to watching both new products enter the market!