This week, I had the pleasure of attending a "Business Intelligence -- Best Practices" workshop put on by the Center for Information Management Studies at Babson College. CIMS sponsors periodic workshops and high-quality project management training. This week's presenters were Henry Morris, an industry analyst from IDG, and Dave Weldon, a Vice President of IT from Iron Mountain. Mr. Morris presented an industry overview, and Mr. Weldon a case study about Iron Mountain's successful BI implementation.
A few themes came up that validated things we've been thinking (and blogging about) at KMA:
- getting the data right is 70% of the effort of a typical BI initiative: activities like scrubbing, de-duping, normalizing, etc., while tedious and time-consuming, are critical to success.
- collaboration and BI are growing together via tools like event monitoring and alerts, process-driven analytics, etc.
- dashboard tools are getting richer and easier to use. Heat maps, and geographical overlays of data resonate much more with business users than pivot tables, which few know how to use.
- the integration of search and BI is critical in the next wave of BI success. Users are used to finding what they want by typing a few keywords into a window and selecting from a prioritized list of results.
- "directionally correct" is often good enough. Data that is good enough to facilitate a timely and correct business decision is the goal, not accuracy to 3 decimal places. Remember General Patton: "I'd rather have a good plan befre the battle starts than a perfect one after it's over."
- while dashboards, scorecards, mobile BI, search and process integration are all trends that are worth keeping an eye on, it should be noted that the successful BI project (over years at a multi-billion dollar company) we talked about basically comprised a well-organized, frequently refreshed, library of key reports. Keep it simple.
A few new ideas were also of interest:
- a product called ClearForest, text-mining categorization software, takes the unstructured data in "comment" fields, finds trends in keyword occurrences, and organizes it into reports findable via search engines (e.g., a comment field in a warranty claim form continues to contain the word "manifold" -- this would give users the ability to observe trends like this and respond)
- the logical next step from there in text-mining categorization, a product like CallMiner's Eureka, capturing and converting speech (e.g., in an inbound call center) and related metadata (tempo, stress, pauses) to provide content that can be parsed and analyzed
One other observation: hardly an available seat in the house. Very high interest levels and demand for this workshop, very popular topic.
Recent Comments