How to: turn data into visualizations that inform decisions

This is the second blog in a series on data visualization. Read the first blog in the series for an introduction on the subject.

What I’ve found particularly interesting while working with clients to produce data visualizations is the lack of clarity around which information needs to be presented. Many clients have embraced data collection and are making use of the charts feature in Excel to create data visualizations. PowerPoint presentations are ballooning as visualizations are inserted to showcase all of the data being collected and the various explorations of data relationships. Having accessible and reliable data is certainly a necessity, and no small accomplishment, but it doesn’t resolve the issue that at the end of the presentation the viewers are left with unanswered questions. How are we doing? Where should we focus next?

This is an interesting dichotomy. We have a data overload and an information drought. You could say that the expression “data-informed decision” is a misnomer. We need data to create information, which we turn into knowledge for decision-making. Unfortunately, without this knowledge, there is a risk of forming decisions based on cognitive biases. These biases, as described by Max Bazerman in the book “Judgment in Managerial Decision Making”, are barriers that we are not aware of and that blur our rational decision-making behaviour.

Imagine a situation in which executives are preparing yearly investment decisions. They are provided with pages of documentation, or perhaps a bloated PowerPoint deck. The executives review the material and, according to Miller’s Law, retain between 5-9 chunks of information in their memory. We’ve heard our clients describe these situations in which investment decision meetings are dominated by the loudest voice in the room, the one who ultimately sways the votes. While well intentioned, we end up with uninformed decisions that are actually based on few data points, even when a glut of data is presented.

This scenario sheds some light onto the value of data visualizations that are carefully designed, edited and grouped into layouts called dashboards. The dashboards help with trade-off conversations because they allow the user to extract relevant information as required.

But who is equipped to put these useful dashboards together? Business intelligence officers and data analysts understand the categories and types of data that exist. They know how to extract the data and how to conduct comparisons to identify trends and outliers. Unfortunately, this type of exploratory analysis simply leads to data overload – often in the form of those bloated PowerPoint decks. Data visualization designers know how to create visualizations that layer data dimensions. This helps to curb the creation of visualizations, but designers are not well positioned to curate the selection of visualizations. The choice of visualizations should reflect the business strategy and requires an in-depth understanding of executive-level concerns. Simply put, effective dashboards not only demand an understanding of data analysis and design, but also of business strategy – a combination of skills that is rarely found in one resource.

The traditional approach to sharing data is from the bottom-up. Available data is gathered, put into a report and shared for review. This approach is not ideal for supporting informed decision-making. Instead, a top-down approach is needed. This approach starts with asking the audience or user key questions: What are your priorities? What decisions do you need to make to drive action? What data will provide you with the knowledge to make these informed decisions? How can the data be presented to enable your analysis and decisions?

Bottom-up and Top-down Approaches

We recommend sharing data through dashboards that are designed around a framework. For example, a management dashboard intended for monitoring performance could use a balanced scorecard framework. This framework was designed to give managers a holistic view of how they are doing. It incorporates a selection of lead and lag measures that reflect a variety of perspectives, but the number of indicators are limited so as not to overwhelm. The balanced scorecard has successfully guided many organizations away from managing their businesses by “looking in the rear-view mirror”, namely by exclusively tracking financial measures, which are lag measures.

With our clients, we use an adapted version of the balanced scorecard, one that works within the government context. This framework helps guide the discussion around priorities, questions, actions and data. We then design a selection of visualizations that communicate the agreed upon measures. We pair new data with historical data to establish baselines, uncover trends, and layer in targets where possible. Only the most relevant visualizations are brought together into a dashboard.

A dashboard is a business tool with an intended audience. The framework you select to develop the dashboard depends on the end users and the purpose the dashboard needs to serve. Once the framework is in place, we recommend adhering to user experience design principles and following an iterative approach, but this is a topic for another day!


If you’d like to learn more about data visualizations, read the first blog in this series.


Leave us a comment: * Your information is never shared