End User Segmentation
As suggested earlier, there are user types delineated by application requirements as well as by skill. If we look at the users in light of their application requirements, we normally see 3 distinct types:
- Traditional query and reporting
- OLAP
- Data Mining
A pyramid of users by BI application type typically looks like the one depicted in figure 6-1.
Figure 6-1 End-user segmentation by product type
The majority of your BI users will fall in the ‘traditional’ query and reporting area. This has nothing to do with their level of skill or their position within the enterprise. The most important user of all may receive one small report once a month. This report may be responsible for the majority of territory assignments and critical sales efforts throughout the company. This user may have minimal computer skills and has no involvement in the creation or maintenance of the output.
A common error in the BI space is to assign business value to the end user’s skills in handling a tool. No one is negating or minimizing the value a power-user may provide, but there is no correlation between their ability to perform technical work and the impact upon the business.
The majority of BI users identified by degree of skill and involvement in the processes will be casual users at best. The profile of the numbers users by technical skills may appear as shown in figure 6-2.
Figure 6-2 End-user segmentation by technical skills
Note that we see a large number of users who have little skill or involvement in the technical end of BI. We also have a large number of users who inter-operate with BI applications at the query and reporting level and not in the more sophisticated types such as Data Mining.
I receive copies of reports and articles regarding many facets of BI including a recent one from a major industry council discussing end user segmentation. The research seemed sound and the conclusions were solid. They suggested that most corporations would have a far larger population of passive BI tools users than they will have active, probing ones. I don’t think a large, complex survey was required to come to this conclusion.
User Segmentation Caveat
The majority of user segmentation surveys I have seen and been a part of may make some attempt to quantify the users by application type, by skill level, and to some degree by processing required. However, seldom does one see a matrix or survey of the business impact of BI solutions by key business impact.
One thing that can be learned from IT is the concept of the ‘run book’. Competent organizations keep a set of documentation regarding jobs that are run on a regular basis. In the days gone by of the Information Center, many ICs kept very accurate records of the BI objects they created or were involved in.
The majority of BI objects created and used today have surprisingly little documentation and information about them recorded anywhere. We seem to constantly hear about metadata today. Metadata capture and interoperation among tools is essential to complex processes such as the construction of a data warehouse. However, does it not seem strange that we would be so careful to track what we built and how we built it but not how we use it?
What information should be collected about our end users? Here is a partial list of the ‘user metadata’. In my appendix I offer a series of checklists for various aspects of BI solutions. For now I would suggest:
- Functional BI category: Consumer, Casual & Ad-Hoc, Heavy Analyst
- Departmental and business information.
- Decision making level within the enterprise.
- BI application and processing requirements (Query, OLAP, Data Mining).
- What tools and data map to these requirements?
- Skill level profile (casual, ad-hoc, and analyst).
- If a provider of BI output, what key output do they provide?
- Business impact of their output if they create BI deliverables.
- Their backup and plan for skills transfer if they move on or get promoted.
- If a creator, whom do they support?
- How and where will they document their BI activities?
|
