Manage Learn to apply best practices and optimize your operations.

8 principles and best practices for good BI dashboard design

BI dashboards are a key tool for distributing data to business users. Here's advice on how to design effective dashboards that meet user needs for information.

Business intelligence dashboards display KPIs and other data for business executives, managers and workers in a visual interface. Good BI dashboard design starts by thinking about UX, as well as the data needs of users and the overall goals of the business. It's not just about presenting numbers, but also figuring out what to draw attention to and how to do so effectively.

"Analytics is only powerful if it drives action," said Penny Wand, a director in the technology practice at IT and business consultancy West Monroe Partners. BI dashboards can help with that and "are growing in popularity because of a business need to increase the value of data assets combined with the fact that most people are visual learners," Wand said.

She added that well-designed dashboards with informative data visualizations can also increase curiosity about data among business users and spread data literacy more broadly in organizations -- for example, by providing operational feedback and guidance to frontline workers.

Here are eight principles and best practices for designing effective dashboards and deploying them as part of business intelligence initiatives to provide views of revenue, product sales, orders, profits and other business metrics.

1. Test dashboard designs early and often

Good BI dashboard design requires exploring which way of representing data makes the most sense to users. Just because something looks nice or makes sense to a skilled BI analyst doesn't mean it will work for the people in the business trenches. Wand said she sees BI teams in many companies spend too much time focusing on the perfect design before getting around to testing it with the intended users.

A good practice, she advised, is to use the dashboard creation process to inform the final design. That involves working with end users in a cycle of visual data exploration and design iteration to see what keeps them motivated to use a dashboard.

Sample Tableau dashboard
Sales KPIs displayed on a sample Tableau mobile dashboard

2. Provide context and direction to dashboard users

Many dashboards don't provide enough context or direction to users. As a result, it isn't clear to people that the information being presented is relevant to their jobs, and they may view a dashboard as noise instead of a useful tool.

"Dashboards are both a sore spot for users and the most important feature you can provide," said Travis Rehl, senior director of product at CloudCheckr, a cloud management platform vendor. BI teams need to craft dashboards for specific user types, Rehl said. Doing so "greatly increases the usefulness of the dashboard and makes it the landing page of choice for individuals," he noted.

It's also beneficial to think about how to help users start their search for data or get information on specific business issues, rather than trying to make a dashboard the be-all and end-all of their data analysis needs, according to Rehl.

For example, good dashboard design should consider what users will want to look into at the start of the workday, he said. The first questions people have can be as simple as, "How am I doing today?" or "Did anything important happen yesterday?" A dashboard should make it easy for them to get that information. "Regardless of the data you're trying to convey, you need to have [entry points] that invite the user to explore your data and find their answers," Rehl said.

3. Remove the data cruft from dashboards

Don't overdo the data you include in a dashboard. Providing too many KPIs and other metrics complicates the analysis process and often leads to dashboard fatigue for users, Rehl warned. "If your target audience is an executive who cares about particular KPIs, only show the relevant data specific to that KPI."

Rehl works with his team at CloudCheckr to exercise restraint in adding content to the dashboards that are built into its platform. There are literally hundreds of ways they could show relevant but not necessarily important data to users, he said. "The key is to understand how our customers need to use this data and design to anticipate their needs."

Soumya Bijjal, director of product management at open source data infrastructure platform provider Aiven, made a similar point. "By designing dashboards to understand each user and present only the necessary information, companies will get a lot more out of their data," she said.

4. Think through the user experience

Dashboard designers sometimes treat all data as equal, Bijjal said. But differences in presentation and analysis capabilities can have a huge impact on how the data in a dashboard is manipulated and used, she added.

For example, in many cases, a user will begin by searching for a specific answer. A valuable dashboard should bring the user to that answer but also offer views of larger trends and provide drill-down data analysis features, Bijjal said.

"Great dashboards are also individualized and customizable, allowing different teams to craft different [versions] for their own purposes," she said. That helps address a common challenge: Dashboards often have multiple audiences. It's hard to provide value to all of them in a one-size-fits-all design.

5. Tie BI dashboards into business processes

Dashboards generally are thought of as informational in nature. Now, though, enterprises can also use them as an analytics interface in operational systems and processes.

"Businesses are transcending traditional BI by infusing what we call data experiences into day-to-day tools and workflows, using the abundance of available data for closing the gap between discovering insights and taking action," said Pedro Arellano, head of product marketing at Google's Looker data platform unit.

Sample Looker dashboard
A sample Looker dashboard shows data on users, orders and sales.

For example, marketing departments are using curated data sets and dashboards to automatically adjust their bids on online ad placements based on how the ads are performing, Arellano said. Designing BI dashboards for such uses requires thinking about how to simplify the way the data is organized and presented to different users involved in the supported business process, he advised.

6. Establish a dashboard development lifecycle

Good BI dashboard design is an iterative process that can be refined and improved to get the best results, Wand said. She suggested that companies consider creating a methodology akin to the software development lifecycle for designing dashboards, with the following steps:

  1. Identify user requirements, and tap cross-functional expertise to determine the potential value of a dashboard and the availability of required data.
  2. Within a few days, evaluate data quality, and work collaboratively to create a dashboard design and validate its analytical and visual effectiveness.
  3. Use initial feedback from users to determine whether the dashboard adds the expected value or could be modified to do so. If not, drop it, and rethink the plan.
  4. Work with the various participants on a high-level analysis of factors that may have affected the ability to create the dashboard, provide the right data or use it effectively. Identify gaps and ways to minimize the issues.
  5. Take the insights from the review back to the first step, and repeat the cycle, while also communicating them to a broader group within your team.

7. Use the dashboard design process to standardize data

Different departments and even individuals within a department sometimes have different ways of thinking about, and defining, the same data. That has implications not only for how the data is used to drive business decisions, but also what's required to ensure good data governance as part of the BI process that feeds information to dashboards.

Wand recommended that BI teams look to uncover differences in data definitions and business metrics when they design dashboards. That can spark conversations to foster a better cross-functional understanding of data relationships and data quality and consistency issues, she said. It can also help initiate or improve data governance efforts aimed at creating internal data standards and policies.

8. Train end users on how to use dashboards

Don't assume that BI dashboards will be self-explanatory to business users and that no training is required on how to use them. A successful dashboard design project "requires a commitment to addressing the human aspects of technology deployment," Wand said.

You should start, she said, by identifying power users who can spot problems and help others get the most out of new or updated dashboards. It's also important to develop a training and communication plan and ensure that there's a feedback mechanism for users. In addition, Wand suggested launching a pilot program with a small group of users before scaling up a dashboard deployment.

Dig Deeper on Business intelligence dashboards and performance scorecards

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.