30 July 2021
How do you make sure that the data that you collect is structured and actionable? What is actually being indicated by the data you have collected?
We think these are two key questions every organisation should ask themselves when they have started going down the path towards being more data driven.
It is easy to forget that behind every number, graph, and metric there is a human. One can sit for days building complex statistical models explaining exactly how X influences Y and how that in turn leads to Z. It is, however, important to remember that the model might not be something that is relevant to the organisation or department it is sent to. The key to effective analysis is understanding what actually matters. Is the most important metric how many people press on a button everyday, or is it more important to understand why they ended on that particular page to begin with?
All good analysis starts with understanding the person who is going to use it. What are the pain-points in their current workflow? What do they know? And how does that influence the information that they need in order to get a better understanding of their role? I think these are three key questions that everyone should ask the person or people that are supposed to make use of the analysis that you are presenting, whether that be in a keynote presentation, through BI dashboards, or any other means of conveying that information.
When starting out your analysis by using the human resources around you the process of doing the “right” analysis can be streamlined. The chance of having a large and complicated model being unused is reduced, and hopefully customers and/or colleagues will be more inclined to use and learn from what you have done. Start small, and listen to the people around you that are experts in their own fields. Make interviewing and discussing metrics with the user a norm and reduce the time spent doing things that ends up collecting dust in some legacy dashboard down the road.
Humans also matter in the sense that you are creating visuals for humans. When one has gathered a sufficient amount of data and structures/analysed it in a way that provides key-insights (with the help of the people that are supposed to use it) it is time to visualise it. A good visualisation can (and often is) the difference between key-insights being used and iterated on versus collecting dust at some far away server. It is important to live by the statement simplicity is key. You do not want to create complicated visualizations that are hard to understand, but rather ones that intuitively show what you are trying to convey.
Simplicity, however, does not mean simple. Whilst it is supposed to be easy to understand it is generally more difficult to make. Once again it is important to test it on the people around you. If you show someone in the office your visual, would they immediately understand what it is conveying? Are key-metrics highlighted?
No one visual will be perfect to begin with. Only by iterating on your designs after feedback to others can you start walking the path of complicated simplicity. Understand the humans that are going to use your creations and what they need. Understand what they need to know, and after time, ask what they would like to know more about.
Both designing visuals and building good models for analysis is a technical workload. However, the output might not be for the most tech savvy. Understanding how to make technical solutions understandable for the non-technical person whilst also using their knowledge to guide your analysis is what we believe to be the most important factor for anyone in analytics. Make things simple but not dumb, profound but not diffuse.
Let us show you how we work with continuously democratizing data, educating within the human- & data mindset, and helping businesses to truly become human- & data driven.
By Viktor Lingslunde, Data Analyst@Sthlm Strat Lab