Despite all the talk about self-service business intelligence (BI), no one has really delivered a tool that makes it easy for casual users to perform true ad hoc queries. Until now.
Most so-called self-service BI tools are better suited to super users, those tech-savvy business people who gravitate to information technology and become the "go to" people in their departments who create ad hoc reports or analyses for themselves and their colleagues. These tools include semantic layers, lists of data objects that users can drag and drop onto a query panel to generate custom queries, and mashboards, which enable users to drag predefined report parts (e.g., charts, tables, controls) from a widget library onto a dashboard canvas.
Visual discovery tools also provide some measure of ad hoc capabilities to casual users within the context of a published performance dashboard. Performance dashboards enable casual users to drill down on top-level metrics and do some root cause analysis, but not perform unfettered exploration. A well-designed performance dashboard is a sandbox that is large enough to answer 60% to 80% of casual users' questions, yet small enough that they don't get lost.
True Ad Hoc for Casual Users
What about the 20% to 40% of time that casual users want to explore data without guardrails? Today's self-service BI tools are just too difficult for them to use. It's not that executives, managers, and front-line workers are dumb, they simply don't have time to learn how to use self-service BI tools or they forget how to use them since they use the functionality so infrequently. It's a lot easier for a casual user to ask Fred, the super user, to do the analysis for them.
For the past several years, I've argued that the search technology is the only way to empower casual with self-service. My BI Framework 2020 holds a spot for such technology, depicted in the wedge that sits between Business Intelligence and Content Intelligence. (See figure 1.) This technology consists of a Google-like tool that enables casual users to type words into a search box and generate ad hoc queries and reports.
Figure 1. BI Framework 2020
The intersection of Business Intelligence and Content Intelligence gives casual users the ability to explore unstructured and structured content through a keyword search interface.
Several vendors offer products with these capabilities, including Endeca (Lattitude), SAP BusinessObjects (Explorer), and Information Builders (Magnify). Except for SAP BusinessObjects Explorer, which is less a search engine than a visual slicer/dicer of semantic layer fields and data, none have really caught on, which has puzzled me. My sense is that the market just isn't ready for this capability.
EasyAsk: Natural Language Meets Voice Recognition
I recently met with EasyAsk, which has long offered a BI search tool that uses natural language processing (NLP) to understand the intent and meaning of words that users type into a keyword search box. In the use of NLP, they are fairly unique. Their tool is the search engine for Land's End and about 250 other online retailers, but they also offer the tool in the BI space. They believe market events are converging to finally make BI search a reality, and I think they're right.
For one, IBM Watson, which trumped World Jeopardy champions this year, has demonstrated the power of NLP to correctly interpret complex human language. Second, the advent of Apple's Siri for the iPhone 4s is teaching a generation of consumers about the power of voice recognition as a search engine interface. EasyAsk has long boasted an interface to voice recognition software but is currently building a iPhone application as well.
"Siri is teaching people how to search the right way," says Craig Bassin, CEO of EasyAsk. Instead of typing a single word and getting thousands of results, which is the case with most search engines operate, Siri encourages people to submit requests using English phases and sentences. "The user interface of the future will be a big search box in the middle of the screen or a voice recognition system," says Bassin.
This sentiment was echoed by Tim Leonard, chief technology officer of US Xpress, a nationwide shipping company in a recent conversation that I had with him. "BI is going mobile and soon our users will submit queries verbally, by speaking a request into their phones."
Another factor that will help push BI search to the forefront is the NoSQL movement, which uses search-like storage indexes to bring together structured and unstructured data in one system, designed for information processing, reporting, and analysis. These systems are raising expectations that users should be able query both types of data within a single interface. EasyAsk, which is designed to run against SQL databases and product catalogs, now boasts a connector to Hadoop that bridges the worlds of structured and unstructured data.
BI search is inevitable. It's the only way to give casual users ad hoc query and exploration capabilities. As voice recognition, natural language processing, and NoSQL databases go mainstream, BI search will surely join them.
Posted December 16, 2011 11:19 AM
Permalink | 1 Comment |