Storytelling with Analytics and the Processes involved
Imagine you are going to buy an house and stepping into a house under construction. You walk through the house which is only almost done. As you make your way through, you understand the layout of the space. The large room would be the living room; the smaller rooms inside a room would be a closet or the bath. You create a narrative of the space whose form and structure is still very raw. You try and understand all the essential information provided to you, before you make the decision to purchase it or even to just move in.
Essentially, a data analyst strives to do just the same. Create an understanding of the crucial data with a call for action. If critical information are hidden under all the jargon, the critical message is lost unless you innovate so as to captivate the audience. In such a case, the user will be bored and left none the wiser at the end of it. There is a general sense of acceptance when pie charts, graphs, and histograms are presented in front of you. Storytelling brings in the creative aspect that will do just that. One can transform a truly complicated analysis with clever use of charts, graphs and colour. Understanding your tools and presentation of the data in an engaging manner is just as important as all the other steps involved in creating the data.
Nevertheless, this process is easier said than done. The data goes through multiple processes before being polished up and ready for perusal. Starting with understanding the business problem, defining the business goals, getting the right data, modelling the data with right fundamentals and then presenting the problem with a user perspective on time with relevant context is no easy task. It requires clarity as to where to source the information, the nature of the required data and problem that must be solved in the current business scenario. In case of machine learning model, additionally finding the model that best answers question is to be determined by comparing the accuracy of different models. Through this, the most suitable model is chosen. Depending on the business requirements the models are created either in real-time or in a batch basis. The model performance is evaluated with multiple tests. The last step in the deployment of any analytics is the visualisations for user consumption. Visualisation tools such as Power Bi, Tableau, Qlik, Looker, etc that generates dashboards, reports enables the models to be consumed with ease.
As one of our customer, Dany Krivoshey, and who we regard as an Analytics Visionary righty stated:
“Analytics is a tool for better decisions. Keep it simple, motivate action and focus on execution. Eventually, if you are piloting the business, you need to know on time which action to take in case your plane needs to take off, land or avoid turbulence. Imagine, being in a plane during the storm. The signals on dashboards must help you to take the right route to your destination. And it’s what can help to make BI powerful. It is not about more charts and more reports”. Link to his post
This leads us to the next part of the storytelling. Its never about showing more charts or more visualisations. Imagine watching a long movie which had no sense of direction. The user will eventually give up and loose interest. An analyst will have to ensure that the story they are telling can lead to a decision which can be made quickly. It will also be very important to make sure you are story is told in time and consistently. Often this is where companies struggle to generate benefits from analytics.
Author: Jerrin Thomas Co-author: Krithika