The challenge of bringing Business Intelligence (BI) to an organization is daunting with a substantial investment in both time and money. There are a number of excellent books on the steps to implement BI, my favorite being the “The Data Warehouse Toolkit” by Ralph Kimball and Margy Ross. This methodology and others take the reader through a series of steps from project planning, requirements gathering, dimensional and technical designs, etc. However, as an implementer of BI systems over the years there are common phases in every implementation that transcend the popular methodologies. Recognizing these phases is critical to making hard work of implementing BI translate into powerful benefits.
"The dichotomy of wanting to appease the user community by making the system more robust will come in direct conflict with validation requests"
This chart is a visual representation of the phases after go-live. The meaning of “Level of Intensity” will vary based on what is being measured. For example, intensity for “system usage” means the relative overall use of the system, whereas intensity for “Challenge to the BI team” represents how difficult it is to maintain the BI system.
In this phase, there will be heavy usage of the BI system initially, but then it will fall flat. The business is seeing usage by a handful of technically savvy users that have been starved for data. These same users, seeing the promise of BI, will request a large number of new BI enhancements.
• While the usage may be lower than hoped, this is not actually a bad thing. Work with these savvy users; they will be useful in a later phase.
The Re-Training Phase
The challenge of taking users from a one-dimensional reporting perspective to a highly flexible reporting world is daunting. Most users will be intimidated by this new found way of slicing and dicing of data in ways that was previously too time consuming or too difficult to implement. So despite a great training program and high-user expectations prior to go-live users tend to fall back to old ways of data collection.
The key here is recognizing when the use of the system begins to drop, and start to double-down on training. Here’s how to prepare for this moment:
• Have Weekly or Bi-Weekly Lunch and Learn sessions focusing on different topics. Keep it simple.
• Create a Portal of “How To” documents. Even the most trivial functionality in the BI system should come with a “How To” document. One example is how to export a dashboard to Excel. The “How To” should be short and loaded with screen shots.
• Provide all training in production or an environment that mimics production.
• Set up meetings with users to create their first dashboards.
• Implement a BI Champions Program. This is designed for a select group of technically savvy users that have shown a level of interest and proficiency in the Go-Live phase. This is an 8-10 weeks intensive training program that covers all areas of the BI system. The objective is to build highly trained end users that will excel with the BI system for their own purposes and become a mentor for other users. Examples of training courses include the following:
LESSON 1 - Advanced Calculations
LESSON 2 - Advanced Dashboarding
LESSON 3 - Visualization Techniques
LESSON 4 - Advanced Comparisons
LESSON 5 –Enhanced Cube Building
LESSON 6– etc.
• Spare no expense on this re-training program. Repeated exposure to the BI system is the only way to help users develop a comfort level with the system.
• Provide supporting materials following each training session.
As users take advantage of the training and adopt the BI system, two key things will happen:
1. There will be an exponential growth of requests for new BI Enhancements. Most of the requests will be for relatively simple attribute enhancements.
2. Users will call into question the validity of data. The challenge here is many of these validation requests will come from users that had no way to validate the data points prior to the BI system and are now surprised by the results. Users that have been operating with some success on gut or intuition may have their beliefs challenged. Here the BI manager could make a fortune charging for every “That does not seem right” request.
The dichotomy of wanting to appease the user community by making the system more robust will come in direct conflict with validation requests. A laundry list of user validation requests creates distrust of the BI system. Make no mistake this is the most challenging time for the BI team!
• Avoid implementing small changes to the system to solve specific end user requests. “Can you add the county to the Customer Dimension?” “Can you add an item master parameter?” These requests are valid and generally easy to implement, however, the challenge is the BI system must be treated just like any software development platform. New functionality should be included in regular software releases.
• Avoid relying too heavily on the validity of reports from previous systems. It is a mistake to assume that legacy reports are the only way to validate the BI system. It is not unusual in BI implementations to learn that reports that the business has been following for years are incorrect.
• Actively invite, even challenge users to find data discrepancies. Acknowledge the receipt of the error immediately and make identifying the correction of that error, a top priority.
• Over Communicate! User trust of the system will be razor thin… bad news will travel faster than good news. Openly acknowledge errors and corrections as part of a process to bring positive improvements to the system.
• Document validation requests carefully, and segment all requests into categories that make sense to the business. For example, create simple categories like purchasing, inventory, sales, financial, etc. Often data discrepancies will have a pattern and grouping issues together will help identify those patterns.
The Growth Phase
At this stage the hard work of the BI Champions program will lead to a group of powerful users that are taking advantage of the system. Old reporting systems will start to fall into disuse.
• Create hard deadlines to shut down legacy reporting systems.
• Set expectations by frequently communicating a schedule of new functionality.
There is no predictable pattern to the timing of these phases. The timing and intensity of each phase will vary greatly by business. For the BI manager, the lesson here is to recognize the symptoms and provide timely fixes to combat the challenges of each phase. The promise of BI is real with the proper amount of nurturing in the post go-live phases.