In many of the large companies I have worked for the institutional BI system has not always been the primary reporting system used by the knowledge workers. A side effect of this is that many users, especially in Finance, resort to developing their own Excel based solutions.
There are various reasons for this, however, one of the main reasons is that an institutional BI system must meet Business & End-User requirements to a sufficient level that the users choose to use it.
The user acceptance covers the following areas:
Knowledge workers don’t have time to wait around for reports. A simple report should be returned in less than 5 seconds. Users may be happier to tolerate a longer query time if the query solves complex business problems.
Are the client tools easy to use? Is there any ambiguity in the cube that the users need to be aware of? The higher the dimensions, measures, hierarchies, levels etc… the harder a cube is to use.
The key expectation from a BI system is that the system provides the information to the user that enables them to; support the decision, provide the right insight, improve performance etc.. It’s important users have access to the data they need regardless of size, or source system.
BI projects can be very complex and time consuming, hence we hardly ever see a BI project deliver all of the internal business data. There is usually a functional focus related to the promises of the original business case.
In highly dynamic organisations the business needs change all of the time. Especially when there are senior leadership changes. Knowledge workers are creative people and in time they will go back to creating Excel worksheets if they need to.
The image below elegantly sums this up.
With self-service BI the knowledge workers are able to mash-up data that exists in many different sources.
In some organisations the business users do not know what they need until they see something. The self-service reports provide no better way to engage with the business to capture their requirements.
A frequent use case is taking data from the BI data marts and enriching it with other data that exists in a source system that wasn’t part of the original BI scope.
The speed that models such as this can be built is days rather than weeks/months. The output can be shared on the Intranet. If these self-service reports prove to be popular they can be fed into the requirements for the next phase of the institutional BI development.
In this sense, the self-service reports are used to prototype reports and provide many elements of the data discovery. And at the same time business needs are met without going back into Excel Hell.