BI Adoption is similar to change. Any BI initiative has resistance from users in embracing the change immediately. The image below classifies different data delivery mechanisms based on user groups. End consumers of data typically look for two key items in a report. Every aspect or the conditions to be met for user to accept a report/dashboard can be classified into two broad categories Precision and Interpretation.
- Precision deals with the accuracy of data, wherein the business definition of KPI and calculation logic should be consistent. This facet is predominantly tackled in the back end (ETL, Modeling, Querying)
- Interpretation is related to the ease of which information is conveyed to end user. This is an art intertwined with technology, where the design plays an important role. Representing data in visual form is one part, whereas ensuring the right graphic widget conveys the right information is critical.
Dashboard designs follow two broad categories
Wide Scoped – Generally these dashboards have several KPI’s incorporated within a single view. In terms of look and feel, they are similar to info graphics to some extent. In short, these type of dashboards can be visualized as ones that contain a large set of KPI’s with minimal number of views/tabs and options to select values. Another angle to view this design method is to incorporate a large set of measures and restricted set of dimensions.
Specialized – Focus here is on a small set of KPI’s and provide more features for analysis. Users typically interact with such dashboards for different analysis scenarios. Each view or page focuses on minimal set if UI elements. This can be viewed as a dashboard with small set of measures and wide range of dimensions.
Although all users are decision makers, if looked closely, the user base of dashboards can be easily bifurcated. Groups can be seen as data consumers and data analyzers, but each user will wear a different hat based on situation context. Wide Scoped dashboard are better suited for data consumers and Specialized ones will fit data analysts.
BI projects follow either a Top-Down or Bottom-Up approach for implementation. In this context, a Top-Down approach focuses on building a Wide Scope dashboard that will appeal to a wide range of users and slowly accommodate specialized ones. Bottom Up approach is the exact opposite and seeks to deliver Specialized dashboards to small user group first slowly expand the scope to build other.
So, What is the best approach to adopt while delivering a dashboard initiative?
Top Down approach has its own set of advantages, but Bottom Up method is adopted in majority of projects. Advantages of embracing this approach are
- Structure of this approach provides inherent flexibility with development cycle
- Leeway available for experimentation
- Better suited for Agile methodology of design and delivery
- Targeting a small segment of users for requirement scoping and delivery is easier
- Development team gains efficiency in delivery as project expands in scope
- Does not seek to implement a “one size fits all” concept
This is not an exhaustive list, but a summary of advantages seen by me personally during my career.
NOTE: It is better to start any dashboard initiative after Precision of Data aspect of the project is complete. If open questions regarding data quality exists, the same issues will percolate to final dashboards.