Everyone loves a good dashboard. To be perfectly honest: dashboards can be stylish, dynamic, and inviting. And we have all heard many times that products need dashboards to bring that “cool factor.” But are dashboarding solutions really that useful?
The term of art used is “visual analytics”. Dashboards allow someone to visually understand their data by leveraging human perception instead of cognition to process the relationship between a bunch of raw numbers. The emphasis on visual analytics has spawned an entire industry vertical of dashboards and Business Intelligence (BI) tools. But while these tools have helped users visualize their data, they often seem to feel lacking when used in practice. They seem to just fall short in solving real-world problems.
This article will focus on how BI-tools have limited utility for empowering knowledge workers to take actions. As a result, we need to start thinking about solutions that help us create products faster; where any visualization of data are there to provide context, and is in service of helping an end user to take actions.
But if you ask end-users of BI solutions, why do they dislike their visual analytics tools, they cite the inflexibility and complexity of use (even when they know what it is they want to do), or the difficulty in setting up a new BI project. And while those are all valid concerns, they do seem to run surprisingly close to the adage attributed (or misattributed) to Henry Ford: “If you ask people how to improve their transportation, they will ask for a faster horse”.
And this is what makes finding an effective dashboard solution so hard. When you only have a hammer (reporting tools), every problem looks like a nail. That hammer will never be the tool to deliver what they need.
So given the frustration, do teams really need a simpler/better reporting solution or is the problem really something else?
So let’s take a step back… why do we need to understand our data? Daniel Rosenberg, design professor at SJSU states in his new book “UX MAGIC” that:
A dashboard visualizes summary information for a family of related objects. The information selected for placement in a dashboard must support decision making. If no decision is facilitated by the dashboard, then it is just artwork to impress executives, not real [Interaction Design].
Clearly, visualizations and BI are not an end in themselves. The GOAL of every user workflow that involves visual analytics is making a decision and then taking action without having to use a separate tool thus doing something with the new understanding, context, or conclusion. The visualizations are needed in service to the end-user learning or knowing what steps to take (and identifying which actions are required). Any attempt at visualization must be in service to supporting that action selection and decision. If you reach that decision to impact your business via any BI dashboard today, you must use one or many other tools to affect that change.
Example Use Case
We all have received that call from our bank, where there is a suspicious charge on our credit card… These calls come from the Financial Fraud Detection team, who monitor suspicious activity and take actions to respond.
The job responsibilities and workflows that they need to accomplish consist of 3 main activities that result in taking action:
- [context] monitor holistic views of the bank and its accounts
- [understand] Identify compromised or suspicious activities/account(s) ASAP
- [understand] Drill in to understand the context
- [take action] Immediately take actions to remediate suspicious activity…in real-time
Notice that, like all applications, the monitoring and visualization-heavy workflows are only there in service to enable the Fraud Analyst to take an action. The use of visualizations (over a large and complex data table) is to speed up the analyst’s understanding by switching the human memory burden from cognition (which is slow) to visual perception (which is lighting fast) and get him/her to take appropriate action rapidly and accurately.
In this example, if all the Fraud Analyst had was a written report on the data at hand, the effort required to drill in, identify the problem and remediate it would be extremely difficult. The desired action would be separated from the context. Success would depend on the chance that the analyst had exactly the right slice of data in front of them at this particular instant in time. Such a report (or such a constrained dashboard) would not provide the right context. Beyond that instant, it is also unlikely that the report/dashboard would seamlessly fit in the workflows needed to come to an effective conclusion.
Thus we can see that “understanding a problem and then taking the right action” is the edge of BI provides. It is the first step in providing a useful application. The goal of such an application is to make an ACTIONABLE system. Any B2B product you buy will have visualizations, dashboards, or other visual workflows to help inform the end-user and guide them to take a step, make an action, move the needle.
And this is why team after team is frustrated with BI and dashboarding solutions. Whether or not the users can articulate their underlying pain point, they are looking for the application to support them in doing something actionable. Visualizations and dashboards are hit-and-miss. If you are lucky, the action required is extremely light-weight. But just as easily they can confuse the user by throwing data together without providing a clear and simple actionable step to resolve the discovery. The need for visualizations is the symptom of the problem (data is very complex), not the cure (diagnosing a problem and coming up with an effective action/solution). The cure is to identify the problem that fits the domain of the business, to determine a solution that concurrently supports both the data and the business workflow, and finally to take actions that solve the problem and improve the business — Is it any wonder that the end-user often cannot articulate their frustration? They have been given a hammer so to them every problem looks like a nail. They need an entire different class of easy to use and affordable tool that does not yet exist until now.
Cover Image by Edho Pratama on Unsplash