Creating a useful business dashboard is an mix of business understanding and technical wizardry. Usually the two of these things don't mix very well. When developing reporting solutions some analysis could lead to more analysis when an interesting finding comes up.

Final delivery is often different to what was the initial ask. On the technical side, things are never perfect. There's always that little bit of improvement that can be done. A report that can look better or a process that can be made more efficient. Nevertheless, there are certain things you can do in order to make your dashboards align with your business objectives.

A dashboard is a visual display containing all the information necessary to achieve particular business objectives of an enterprise. You can think of it as screen in the control center of the USS Enterprise. The crew have everything available to successfully complete the current objective. Using the relevant information at hand and the guile from the main characters, everyone is happy at the end of the episode.

Think of building a dashboard as developing a software product. You will start with a set of requirements, explore the landscape, build and test the solution and hand it over to the users. There are things you can do at each of these stages to deliver a better product at the end.

Keep in mind:

  1. You don't know if you have the data available for the required reports.
  2. When you start refining the idea, it may lead you down multiple paths that you didn't know existed in the beginning.
  3. Life is all about technical compromise.

One important thing to note is that in most cases, you never end up with the dashboard you set out to build in the first place. The more you work on it, the more refined the knowledge of your solution becomes. The more you think about it, the more you start to see new possibilities.
Because of the constant shift in requirements, a good methodology for creating dashboards is Agile. With it's emphasis on creating small shippable solutions, constant customer focus and the requirement to rework when the requirement changes, it perfectly mirrors the dashboarding process.


Gathering Requirements

One of the fundamental truths in building things is the further along you are in developing the solution, the harder it becomes to change course so it is important to understand what is needed as best as possible before you embark in creating a solution. You must fight the urge to just start building stuff and do the proper steps in analysis before you begin.

Try and work out:

  1. As much detail as possible about the business expectations of the solution you are building.
  2. What falls into scope and what is outside scope.
  3. What are the objectives of the dashboarding exercise?
  4. What indicators determine success of failure of the exercise?
  5. Who are the end users and the level of granularity of the reports?
  6. What are the dimensions? A dimension is a category used to summarize business data. E.g. region, business unit, branch manager.
  7. What are the filters that you can apply? A filter is a way of getting a smaller more relevant dataset.

Create a requirements list and communicate this with the other stakeholders. The more people validate the requirements, the more likely the requirements are accurate. It is important that there are proper communication and reporting channels open to all the stakeholders.

Data Sources

Getting to know the data geography of organisations will give you a firm grounding upon which to build dashboards. To build the most relevant solution, you need to have a top down approach to dashboarding. Here, you start with the requirement and then start looking for the data to satisfy it. If the data is not the way you want it, get as close to the ideal dataset as possible for the time and then aim to get it perfect.

Figure out:

  1. What are the relevant data sources. Databases, Spreadsheets, CSV Files, Webservice, ERP
  2. Do we have the ideal dataset or do we have to compromise?
  3. How often is the data refreshed?

Documenting the sources of data will be invaluable if the dashboard starts to spawn requests to make other dashboards. For each data source, document details about data formats, extraction timetables and what fields are relevant. The team could be come back to multiple times if there is more analysis to be done. In a growing team, documenting datasets always puts you on a firm grounding.


Automate, Automate, Automate

A sophisticated dashboard could take data from multiple sources. There is no value in manual processes if they can be automated. There is potential time wasted and errors created at every point human hands touch the process. Even with the best testing and validation in place, it is better to automate everything you can. Whether it is using VBA on Excel or Python's Pandas library, elegant solutions can be built to automate even the most complex processes.

How the dashboard will be distributed to the end users. The time frames could vary from being real time to reporting for month end reviews. Depending on the frequency of distribution and the availability of data, put a timetable in place so all involved will know what to expect.

Overall Design

Due to the iterative nature of the project, the above steps are unlikely to happen sequentially. Also, as each project is different, the focus on each of these areas can change depending on where and how the dashboards are implemented.

Can you:

  1. Separate data from analysis from presentation. A more modular approach to delivery has the benefit of readability and adaptability.
  2. Implement proper error reporting. Although the possibility of errors cannot be completely eliminated, the chance of them happening can be minimized.
  3. Add a how to guide on running the dashboard. This will include the steps to take and what to do if something breaks. The final handover document will contain the guide as well as the other documentation arising from the other steps.

We would really like to hear from you and wish you all the best in your dashboarding endeavours.