If the statistics are boring, then you have got the wrong numbers
Edward R. Tufte
During the period of assessment of what is needed for the future dashboard, it is important to focus on the Goals, not on the means. If people are asked about the dashboard needs, they often think about the appearance of the final product, basing their expectations on the examples that they have seen before (in many cases, poor examples). It is important to shift the focus on what people are trying to accomplish rather than a specific look of the dashboard.
Questions to define a dashboard needs:
- How frequently the information should be updated?
- Who will use the dashboard? Is it for a single person, a single group, or people in several different departments?
- What will the dashboard be used to monitor, and what objectives will it support?
- What questions should be dashboard answer? What actions will be taken in response to these answers?
- What specific items of information should be displayed on the dashboard? What does each of these items tell you, and why is that important? At what level of summary or detail should the information be expressed to provide the quick overview that is needed?
- Which of these items of information are most important for achieving your objectives?
- What are the logical groupings that could be used to organize items of information on the dashboard? In which of these groups does each item belong?
- What are the useful comparisons that will allow you to use these items of information in meaningful context? for instance, if one of the measures that your dashboard displays is revenue, do you have targets or historical data that could also be displayed to make current revenue more meaningful?
People accustomed to receive reports with information that they do not use for anything. However, they are afraid to loose this information… How to encourage removal of these information: “Describe a situation when this information would lead you to do something” or “Give me an example of the actual data that would appear and the action that you would take in response.” If no examples come to the person’s mind, then this information does not belong on the dashboard.
Many dashboards suffer from the problem of lack of context – they present the information that will never inform action because the context that is needed is missing.
Non-quantitative data may need to be present on the dashboard
- top 10 customers
- issues that need to be investigated
- people who need to be contacted
Challenge of the dashboard design:
- make the most important data to stand out from the rest (based on human perception)
- reduce “non-data” pixels
Insisting on cute charts when another means would work better is counterproductive even if everyone seems to be in love with them. The appeal of cuteness will fade quickly; the only thing that will matter is how effectively and efficiently the dashboard communicates needed information.
- Pie chart is more difficult to perceive compared to a bar chart – it is easier for people to compare lengths than area – people are very inaccurate in estimating relations in different areas
- Funnels are also not recommended – it is easier for people to perceive sizes of the funnel parts – it is much easier to compare simple bar chart
- Explanation of a bullet graph
- Do not represent two different data sets on one graph using left and right axis – users will attempt to compare unrelated data sets – use two separate graphs instead
- We should not change type of display type (type of the chart) just for the sake of a visual variety on the dashboard. the goal is to best select the medium that best communicates the information, even if the dashboard consists of the same type of chart
Begin designing the dashboard by sketching
Beware of focus Groups
- Focus groups engage “system one” thinking, often rejecting anything unfamiliar
- When asking to review the dashboard, invoke “system two” – do not show the dashboard and ask “Do you like it?” but rather review the requirements, explain that the dashboard was designed with consideration of human perception, and then show the dashboard. Without allowing too much time to pass, start asking the questions such as “are the groupings of information obvious?” or “are your key metrics being measured adequately?”
To design dashboards that really work, we must focus on the fundamental goal: communication.