How to build an analytics dashboard

Data visualization is one of the most basic features of any business intelligence tool. For most employees, it’ll be the main way that they interact with their BI tool; they won’t be doing any ETL or running their own analysis, but they will be viewing dashboards to monitor their tasks and projects, and they might even create dashboards and visualizations of their own.

Many modern BI tools expect that their users will want to build their own dashboards, even if they don’t have any data experience. These tools, part of a movement called self-service BI, give every user the tools they need to make powerful, effective, useful dashboards.

However, having the tools to build a good dashboard doesn’t automatically mean that the average employee will actually be able to build a good dashboard. Self-service BI tools may offer workers the tools, but they don’t offer a lot of guidance on how to actually design and organize a dashboard for the best results.

Dashboard design and organization is a skill all its own. It requires some graphic design knowledge, but also takes in-depth knowledge of the topic you want the dashboard to track, and at least a little insight into how and why to use different data visualization. A good dashboard can help a project move quicker and boost efficiency, while a bad dashboard can actively harm a team’s data acceptance.

Here are some general guidelines for making dashboards that clearly and effectively get across the information you need them to.

Choose the right metrics

Most dashboard advice focuses on making sure the dashboard is clear and readable, but before a designer builds the dashboard, they have to decide what information to include and what to leave out. It’s very easy to build a well-designed, intuitive dashboard that no one uses because it doesn’t have the data that users need to know.

Only include the most important information

Dashboards should contain the analysis that’s most important for the dashboard’s audience to do their jobs. The metrics that are most important to a dashboard’s topic are called key performance indicators, or KPIs.

For a given topic, there might be hundreds or thousands of different data streams that a designer could choose to include on their dashboard. However, only a few of these metrics are likely to be important. The most important metrics to include on dashboards are ones that the majority of users will need to access frequently. That’s the point of a dashboard—to save users time having to manually access the same data over and over again.

Make it intuitive for the end-user

There’s no upper limit to how many visualizations a designer can include on a dashboard, but if a dashboard has too many visualizations, then it won’t be useful as something that can be quickly scanned for information, even if it does convey a lot of information at once. A good rule of thumb is anywhere from 5 to 25 visualizations, depending on the size of the page and the size of the visualization.

With interactive dashboards, designers can help users access information that normally wouldn’t be included. Using tooltips and thumbnails, designers can present multiple data sets in the same visualization. Users may be able to click on a visualization to drill down, where they can see more information related to that visualization on a sub-dashboard.

Organize the dashboard

Dashboard builders can’t put their visualizations and statistics wherever they feel like. One key way to ensure that a dashboard is clear and readable is making sure that the different components of the dashboard are well-organized.

Prioritize

Dashboard builders need to make sure that the most important information on the dashboard is prioritized, while less important visualizations are de-emphasized. There are a few different ways to do this. One strategy is to place important visualizations in the center or top left of the dashboard, and move less important things to the periphery. A designer might use bright colors for important graphs, and use more muted colors for others. Another might make some graphs and charts physically larger than others to show primacy.

Think like a designer

A good designer will use these instincts to make their dashboards easier to parse naturally, so that viewers don’t have to actively look for the information they need. Instead, they end up looking at that information naturally, following their brain’s basic rules for absorbing information visually.

Designers can use similar tricks to make sure that dashboard users view things in the right order and in the right ways. When they need to look for information related to the visualization they’re looking at, most native English speakers will look to the right. This is because English is read left-to-right, so it feels natural to look to the right to get a continuation on the same topic.

Layout best practices

It also helps to put related visualizations next to each other. It’s best to think about what data a user might want to view after looking at a given visualization, and then putting that visualization close to the original one. Users shouldn’t have to look all around the page to find a visualization; it should be where it feels intuitive to look.

A common strategy for arranging elements on a dashboard starts with a primary visualization in the top left, then related information to that visualization trailing off to the right, and then a new topic or primary visualization on the next ‘line’ of the dashboard. This arrangement strategy mirrors how English speakers read text, which makes it fairly intuitive for the average person to understand.

Use the right visualization for your data

The type of visualization that a dashboard builder uses to convey their data can have a massive effect on how the data is perceived. Not every visualization expresses its information the same way, and understanding the ways in which data visualizations differ is an important element of dashboard design.

Use summary numbers to convey KPIs

Some dashboards use summary numbers to present simple information. A manager that wants to know a given KPI might not want to know how it compares to other metrics or how it’s changed over time; they just want to know the current number for that metric. In this situation, a dashboard can just display that number instead of using some fancy visualization to abstract it.

Most business intelligence tools have hundreds of charts and visualizations to choose from. It can be tempting to break out complex visualizations as a beginner dashboard builder, but in most cases, the more basic a visualization is, the more useful it will be to viewers.
 

 

Why is design so important?

Dashboards are an essential part of any good business intelligence tool. They’re the most efficient way to connect the average user with the data analysis they need to do their job.

In the past, most dashboards were built by data scientists who often didn’t know or care about the end user’s needs. This led to a lot of dashboards that weren’t particularly useful. With modern, self-service BI, users have the opportunity to build their own dashboards. However, to make good dashboards, users need to understand the best ways to convey information visually, so that their dashboards can be understood quickly and clearly.

Good dashboard design boils down to just a handful of key ideas. First, users need to figure out which of their data streams they actually want to include on the dashboard. Dashboards have limited space, so designers can’t include all the information everyone might want or need at once.

Once they’ve figured out what should be on the dashboard, dashboard builders need to organize the dashboard for usability and clarity. The most important data should be easy to find and understand, and there should be logical connections between adjacent visualizations. The whole design should make sense, and be intuitive to use.

After they organize their dashboard, builders have to choose their visualizations. Each type of visualization has strengths and weaknesses; designers need to use the right visualization strategies for their data so that users can actually understand their analysis.

A poorly designed dashboard can wreck a project, while a well-designed dashboard can boost data literacy and drive new kinds of insight. Dashboard builders need to understand the hows and whys of dashboard design, so that they can leverage the power of dashboards to find insight.

Conclusion

By following best practices and being intentional with design, you can ensure that the end-users of your BI dashboards can gain as many insights as possible. While dashboard design may not come easily to many data-centric BI users, it’s a critical piece in creating a polished final product for the data consumer. Leveraging the concepts we talked about in this article will help you become a better BI developer for your company.