User requirements

The initial briefing might be one of the most important steps to avoid getting lost in all the possibilities, and actually reducing complexity with your dashboard. Focusing merely on creating delightful graphics and interactive interfaces may help to impress users, but if your dashboard lacks functionality it is ineffective and not likely to be used on a regular basis. Therefore, your thinking should start with the purpose of your dashboard and the corresponding user requirements. Once you have clarified these points, the design of your dashboard will more or less fall into place.

While there are many ways to slice the cake, the following four leading questions for clarifying the requirements work quite well in our experience:

Who are the users of your dashboard? How often will they use it, what are their daily routines, what goals do they try to achieve and what kind of decisions will be based on the information? Keep in mind that dashboards should always make someone’s life easier; – this is why the user comes first.
What are their individual requirements? Will your dashboard have to include different user roles and how do they differ? What are the right permission levels for each of them to access, analyse and export the data?
What information is relevant for the users? What are the relevant KPIs for each of your user groups? What other data sources need to be included to contextualize these KPIs (e.g. CRM, ERP, …)?
What is the company’s strategy? How does the dashboard help users focus on the right things and align their actions with the overall goals?

User roles and their corresponding dashboard

Now that you have an overview of the general requirements, let’s have a closer look at typical user roles and their corresponding dashboards. The following categories may help you to increase the focus of your dashboard.

Operational dashboards are used on a daily basis by operational teams. They are needed to make well-informed decisions and very often work with real-time data. Their analytical capability is limited, they only stress the most important KPIs to keep the business going. An example would be a dashboard where advertisers can check the audience fit of their ads in real-time and optimise the campaign accordingly. As the data are reported in real-time, operators have a direct feedback loop for their actions.

Analytical dashboards are most often used by researchers or business analysts. The main focus of analytical dashboards isn’t immediate action but rather exploring data, identifying opportunities, or investigating problems. Very often, the underlying data contains a lot of different variables but isn’t updated too often. Such dashboards allow their users to mine patterns, trends and anomalies in the data and, therefore, are equipped with interactive statistical features. Once you have found relevant insights, these dashboards allow you to share your findings with others (e.g. by exporting them). These dashboards are most often used in large projects, especially international or tracking studies.

Strategic dashboards are typically used by the general management to track progress on general business goals. They do not necessarily have to operate with real-time data but should provide a comprehensive overview of all relevant business areas. Therefore, they typically include different data sources that cover all perspectives, e.g. CRM data, client and employee satisfaction, financial data, etc. Especially for companies with branch offices and subsidiaries, strategic dashboards allow benchmarking single profit centers against the overall business performance and grant different users individual access levels and benchmarks (e.g. General Management, Regional Managers, Store Managers).

These categories are not set in stone, of course. There are many cases where a combination of different approaches makes perfect sense. They should rather be seen as a useful tool to better understand the requirements of your specific case.

Relevant data sources

Let’s discuss the data source(s) for a moment. Dashboards can work with all types of data. In the most common case, this will be structured data (e.g. from quantitative surveys), but it can also include unstructured data that has been processed previously (e.g. from qualitative online diaries). Our dashboard solution allows you to perform the typical procedures of data preparation such as coding, text analysis, or weighting.

An important thing to consider is how often the data source will be updated. For Ad hoc studies, the data may be uploaded manually when launching the dashboard. With tracking studies, repeated updates are required. Auto FTP or an auto file reader can do the trick in these cases. Some dashboards may also demand data in real-time. This can be done via a live API.

Finally, different data sources can be combined in one dashboard. We have already seen that some dashboards may require more than one data source to contain all relevant information in one place. This can be achieved by showing information from different data sources on one screen without integrating the data (e.g. customer satisfaction next to employee satisfaction), but also by linking the data sources via a shared variable (e.g. sales of ice cream and weather information at different moments of time).

Once you have established an overview of all relevant data sources and the available variables, you need to start working on the information hierarchy of your dashboard.

Information hierarchy

As discussed, dashboards help to reduce the complexity of all available information. Therefore, dashboards should always allow its users to focus on the most relevant aspects of an issue. A key principle for designing dashboards is making it as simple as possible for the users.

This brings us to a fundamental psychological principle we need to keep in mind when designing dashboards: If too much information is shown at once, the amount and complexity of information can overwhelm the user and actually inhibit decision-making. This phenomenon is known as analysis paralysis.

We’d like to present two basic techniques that help users to stay focused. The first one comes from journalism and is known as the inverted pyramid or BLUF (bottom line up front): try to communicate the most noteworthy information at the beginning, continue with significant details or trends, and put general background information and other details to the “fine print” of your dashboard.

The second technique comes from UX Design and is known as “Progressive Disclosure”. Don’t show everything at once but make use of interactive elements to guide the user through your data, for example with the help of visual hierarchies, tabs, accordions, or filters. Start with the most relevant findings on the initial display, keep unnecessary findings hidden for as long as possible, and reveal these details progressively in subsequent displays. This helps users to focus on one thing at a time while still having access to all information.

With this in mind, you should approach the data you want to visualise in your dashboard. Try to put one metric into the centre of the user’s attention (e.g. general customer satisfaction) and make all other metrics relatable to it. What are the drivers of your key metric (e.g. satisfaction with price, with quality, …)? How do your customer segments differ (e.g. demographics)? How does this metric change over time?

Establishing an information hierarchy will help you to define the sitemap of your dashboard and what visual elements and functions will be needed on each display to convey the relevant information.

Data visualisation

The reason why visualisations have become so popular in recent times is that they turn abstract and complex data into a tangible and relatable experience. Once again, they serve the general purpose of making it easier for the user. A small number of simple design principles can help to make your data even more digestible. Let’s have a look at them.

Very often, the design of dashboards is driven by making them as visually appealing as possible. In general, there is nothing wrong with that, as it can help motivate users to use your dashboard on a regular basis. However, all charts still have to communicate the relevant information and a design decision should never stand in the way of communicating data successfully. This principle is broadly known as “form follows function”.

An example: it might be tempting to group several doughnut charts next to each other because their visual similarity creates a consistent appearance. And, in fact, this would be a sensible design decision if all variables contained only a few characteristics. However, the legibility of a doughnut chart suffers the more values it contains. It becomes harder to identify the mode and to compare the relative frequencies of single values with each other. In this case, it would probably be better to select a different chart type to communicate more effectively.

Removing all unnecessary design elements can help your user to better focus on the relevant data. This principle is known as a high data-ink-ratio among data visualisers. This concept may seem to be a bit abstract at first, but it allows you to assess how clear and simple the design of your dashboard really is. If you think of all the pixels of your visualisation, the amount of pixels needed for displaying the actual data (data-ink: lines, areas, axis, data labels) should outweigh the amount of pixels that are not needed for displaying the data (non-data ink: background grid lines, 3D effects, shadows, redundant labels). This will reduce the amount of distracting design elements and help the user to stay more focused.

Our final recommendation for designing better dashboards is don’t forget the user. Many users of your dashboard will only rely on the information provided by the dashboard without having someone to give guidance in interpreting and contextualising the insights. Therefore, some of the most relevant elements may actually be verbal descriptions that help users to contextualise the information:

• Use headlines, explanations or definitions of variables.
• Give as much (business) context as possible to your numbers.
• Be consistent in the naming conventions of variables and KPIs to eliminate a source of confusion.

Maintaining the dashboard

Last, but not least, don’t leave your dashboard once it’s ready. You should schedule regular revisions. In many cases, the data behind your dashboard will have to be updated over the course of time, anyways. This would also be a good moment to optimise some of its functionalities. Collect feedback from the dashboards users to see what is useful for them and what can be improved. In this way, you can further increase the business impact of your dashboard.

Bottom line

We believe that communicating insights should have the same quality standards as collecting the underlying data. Remember these important steps to get started with your dashboard:

1. Start with the purpose of your dashboard and the corresponding user requirements
2. Use relevant data sources for your dashboard
3. Make it as simple as possible for the users, and guide them through your data.
4. Establish an information hierarchy
5. Use simple data visualisations to make your data more digestible
6. Keep your dashboard updated and relevant

We hope this guideline has been useful to you. As always, all rules have exceptions, but it’s useful to have some general principles at hand that guide and improve our practice. Conversely, we hope you agree with us that good visualisation cannot make up for poor data quality.