“Just Get It Done!”: Avoiding the Pitfalls of Building Dashboards Part 2/2: Making Sense of Dashboards
by Adam Gertenbach, on Feb 26, 2021 3:43:34 PM
A great dashboard embodies all the same strengths as a great business intelligence platform, growing and adapting with your business, and that foundation requires some planning; like any business function, just because you can do it quickly doesn’t mean you should, and there is a bevy of trade-offs between the tactical and strategic approach to building dashboards.
Effective dashboards account for three core elements which you should be able to clearly define as part of your development process: Premise, People, and Presentation. These elements may be approached progressively, building a roadmap for a formal dashboard proposal ready for evaluation.
The Premise: How does this enable us to act?
Every dashboard should start with a premise; more specifically, the kernel of a dashboard and each of its features should be a valid response to the query “What is the business question being asked?” At least once, we have all probably produced a requested report or spreadsheet without any clear understanding of its desired use – “We need it, we would like to see the numbers laid out, we want to see the data.” But, why? Is there an expectation for insight to leap off the rows and columns of numbers and values unbidden? Or perhaps the data is being approached with an existing bias, a search for a way to organize the numbers to support a preconceived expectation instead of the other way around? The flexibility of the self-service approach to data is undoubtedly important and often critical, but it should be a conscious, deliberate decision and not an accidental byproduct of poor or hurried design.
“If we build this, what action does it enable the business to take?” Before starting any prototyping or design on a dashboard, you should know why you’re building it. Not every business user is also an amazing analyst or designer; many times, ideas or concepts are incomplete and fail to get to the root of a question, or propose data elements and visualizations for the sake of data and visualizations. Effective insight leads to action, and so too should your dashboard. Modern analytical tools can be quite a paradigm shift from the multi-layered spreadsheets some users are familiar with as the output of data repositories, so truly understanding the end game will enable more effective design.
For example, a business user once presented me with a request for a dashboard that produced a large table of information onscreen with over twenty columns of numbers. This so-called “dashboard” automated a process that he had performed previously through manual exports of accounts receivable data from a financial system and aggregation of that data in a spreadsheet; the dashboard, he argued, would save him half a working day every week. “That’s great, but what do you do with that table?” I asked. Turns out, he compared it against the prior week’s table and highlighted the top five deltas for several of the metrics to determine vendors who needed to be contacted. The root business question was not “How can I automate this manual process to get a table of numbers from this financial system?”, it was “What are the biggest increases in accounts receivable compared to last week so that we can perform more efficient collection?” This was a much better business question, as it presented a more focused idea and led to a specific purpose behind the design choices and visualizations ultimately used.
“What is the benefit of the action this insight supports?” An established premise provides a means to measure the potential business value of the project and to assign more objective prioritization. Does this dashboard lead to increased revenue, and if so, how much would we project to gain? Will we see cost savings due to improved efficiency, and what would be the benefit? Is it tangible or non-tangible? If you have ten requests for new dashboards or updates to existing ones and you know what action each premise supports, you can measure its value against the investment of your limited time and resources; a project with an established, defensible ROI should take precedence over those that will produce a less meaningful impact.
Once you understand the business question behind a dashboard, you’re better equipped to evaluate the true scope of the request. Is there already a dashboard that does something similar that this would intersect with? Do we have any related metrics or concepts? Is this really a new dashboard, or should it be an enhancement or upgrade to an existing dashboard? Remember, the goal of a dashboard is to be a centralized hub for information; while a single, monolithic dashboard for an organization is probably not the answer, neither are a hundred fragmented ones. Logically grouping important concepts, related metrics, and practice areas will help you maintain purpose for your users when accessing a dashboard. If you must create a new dashboard, understanding the business question will also permit you to determine the time frame of its perceived value; does the dashboard serve an ongoing purpose, or is there an end condition upon which it can be retired? Periodically reviewing the known premise behind your dashboards can help establish whether they may be decommissioned later once they have served their purpose.
The People: How does this integrate into our daily business?
Once you have a premise, you need to understand your users. Understanding the specific classes of users who will interact with your dashboard will help transform your premise into a more effective, detailed story. We’ve already covered that a dashboard is a centralized hub, but that doesn’t mean that all your users will share the same insight and the same experience as they engage with the information presented; an executive isn’t necessarily going to be interested in the same drill-down, self-service data that your analysts are.
Effective user-centric design practices can be compared to the concept of stories and epics in the agile project methodology. Each class of user has an existing flow to how they perform their role that can act as guidance on how an effective dashboard can be integrated into and enhance that flow. If a dashboard is meant for a finance team to be used by financial analysts, financial managers, and your CFO, how specifically will each type of user benefit by using your dashboard? What information will provide insight that will lead to action? How will they interact with the dashboard on an ongoing basis? The collective needs of the different classes of users, their respective levels of engagement, the relative criticality of their roles in driving action, all help form a more refined vision for the practical implementation of the premise for the dashboard.
The form and function of your dashboard will begin to take shape through these stories to give you a foundation for prototyping and to ensure that your dashboard, once live, will be utilized. This is an easily overlooked aspect of designing dashboards, or any other business system or process, for that matter; no matter how much time and effort and theorized benefit a dashboard will offer your business, its practical benefit is nothing if your users do not use it. Including your users in the design process gives you an opportunity to present your concept to the end consumers in advance, letting you answer questions like “Is it intuitive? Does it make sense? Is it easy to use? Do you gain something actionable from your interaction that you did not or could not get before? If this is something you already did before, is this dashboard sufficiently more efficient to make you use this instead?”
Finally, brainstorming about your user base early during project scoping may even uncover potential groups of users you had not even previously considered. Just because a portion of your organization does not now gain insight from information currently stuck in a silo does not mean they couldn’t or shouldn’t; dashboards are a great opportunity to remove the silos and share information with everyone who may benefit and act on it. Perhaps your financial team would benefit from operational insight, or vice versa? Maybe marketing would be enhanced by understanding the flow of goods through distribution? It is important to keep your project focused and develop in reasonable, realistic phases, but understanding the big picture may enable you to make small changes with a big impact and to set the groundwork for future enhancements and updates for the benefit of your entire organization.
The Presentation: What’s the most effective way to deliver this insight?
Now that you know what insight you’re delivering and you know who can act on it, you must determine the best ways to deliver it. The people-based stories you’ve developed are the foundation to choosing the specific visualizations and media that are appropriate to facilitate efficient consumption of the information. Your dashboard may be more open-ended, encouraging users to engage in self-discovery and self-service based on a series of core KPIs at the forefront, or perhaps it is meant to tell a story through data, guiding users through a predetermined path to drill down into information. Should it be interactive, allowing lots of filters and selections, or is it more hands-off? Will your users be accessing via a laptop? A tablet? A phone? Or perhaps your “dashboard” might really be more beneficial as a fixed report sent as an email to everyone’s inbox each morning? Together, these decisions will influence the presentation of your “dashboard”; you must establish the appropriate balance between guided and self-service analytics based on the premise and the people who will be consuming it. If you are a successful business, there is probably some method behind that success, and it’s important to consider the existing systematization that drives your organization when providing analysis; while you do not want to hinder new, creative avenues to look at your information, you might also want to encourage paths of thought that you know are proven to work.
Presentation also involves the effective use of design and visualizations. Have you ever seen a dashboard so cluttered with content and color that was hard to grasp what it was trying to say? Great dashboards consider the use of space and color, the organization of content, they provide an intuitive flow with information that is quick to grasp. There are probably several different charts or graphs you could use to visualize a given metric, but some are better than others based on the content. Trends over time may lend themselves to line graphs, ranges and outliers may be better illustrated by a box plot, and sometimes the best way to show something is a simple color-coded KPI. If you have a ton of information in a straight table, stop and ask yourself what that table is meant to do. Much like the earlier example about the highest changes in accounts receivable, many people look at a table to answer a specific question; can you answer it better with a more refined visual?
It’s easy to rush into dashboards, deploying them with reckless abandon because they can be quick to assemble, and it’s equally easy to inadequately budget the time and resource investment required to build good dashboards and to maintain structure and standards in a business intelligence environment. However, as with any business infrastructure, this methodology leads to an excess of technical debt that you will have to pay one way or another! Consider the strengths of your platform, encourage strong governance, and remember to build with a premise, build for people, and be conscious of presentation.
Here at Axis, our architects understand the critical foundations behind successful business intelligence solutions and have years of experience working with clients to develop structured, strategic analytic applications. Our visual analytics team offers comprehensive design engagements to help you find the right way to deliver insight to your users, from out-of-the-box dashboards to fully custom solutions, our data science team can help you answer even the questions you didn’t know you could ask your data, and we offer managed service solutions where we can host and maintain your environment for you.
Do you know where you can improve your business? Your data does. Let us know if we can help!
Adam is a Solutions Consultant at Axis Group with years of experience developing custom solutions for data management and business intelligence. He's dedicated to helping clients understand how to deliver measurable business value and design strategies for the long term.