In this short series I take a look at the how User Interface (UI) design is of vital importance to business users and begin with a review of a book that exposes the ugly underbelly of the typical “dashboard”.

One of my longtime passions has been user interface design (the way a screen is laid out for users to see and use in a software program) for two key reasons. The first is that effective visual design can, like art, provide a “relaxing” atmosphere for users to openly understand and “consume” the information they are seeing, and the second is that an effective and functional UI is essential to increase user productivity and acceptance of software.

These two principles are inextricably linked and are particularly difficult to achieve because they are as much art as they are science. I see this topic as far more than an interesting sideline as it underscores so many reasons why certain softwares are well liked and others destested by the user community, even though they perform essentially the same function.

Stephen Few’s book Information Dashboard Design (O’Reilly, 2006) provides a pragmatic introduction to this topic because he articulately dives into how Business Intelligence dashboards are critical but usually flawed in their implementations. At the root of the concepts presented is that information on a page, or a screen, is perceived by our brains in a specific order with built-in evolutionary “filters” that automatically process certain information efficiently.

Position, colors, shapes, shades, and other elements come together to either make it easy or difficult for our minds to understand. The best designs take advantage of our “preattentive” abilities to understand shapes and patterns without having to consciously think about it, therefore making the experience totally intuitive and rapid. As an example, if we see a star or color highlight beside one of many rows it jumps out at us without any brain work. See below:

Sales:          +10%
Cost:            + 2%
Margin:      –  4% *
Turnover: + 5%

But if half of the rows have a star or they are all in bright colors, then the value of the exception is lost – nothing stands out as important. In a poor design we see these typical mistakes:

  • The wrong type of graphs waste space and provides little information (e.g. pie charts are almost never useful);
  • Expensive screen “real estate” (like the top left of a screen) is used for a big logo instead of a meaningful chart; 
  • Gridlines and colors are overused and force us to process the “non-data graphics” (those pixels that have nothing to do with useful information);
  • Unrelated elements are perceived as related because they are shown in the same color or are put too close together (we do eventually realize the lack of relationship but it took our brains some processing time and therefore we are slowed down and distracted). 

All these problems (and many more) lead to delays in the processing of valuable information on a screen, and at worst a total misunderstanding. Usually the effect is a dislike of the interface leading to low user adoption rates – a costly error on the designer’s part. And yet these dashboards are intended to provide key strategic and operational information to allow staff to make critical decisions.

Therefore a failed visual design is a significant problem to any organization. Few goes one to document dozens of examples of how dashboards are presented by (usually) BI vendors and how their own designs show a total lack of understanding of basic principles of effective design.

See an example in figure 1 below of a publicly available dashboard from a popular vendor’s site. As Few explains, he has nothing against these vendors, he simply is pointing out how they generally, to be blunt, don’t “get it”.

Figure 1: These gauges are a big waste of prime screen real-estate

Figure 1: These gauges are a big waste of prime screen real-estate because they offer very little useful information.








He presents interesting new advances in dashboard design such as the bullet chart (his own invention) and the sparkline (invented by Edward R. Tufte), both of which provide more intuitive views of data in a compressed space. After studying these you realize there is a lot to learn in this science and Few is on the right track.

He closes off by going through comprehensive examples of good design samples as compared to equivalent ones that don’t pass muster. Impressively, he is using real-life designs as his examples of what not to do. He isn’t making this stuff up and it hits home.

The lesson: vendors and designers listen up, there is a much much higher bar to attain in designing dashboards and it begins by studying how humans naturally process information.

Rating: **** (4 stars out of 5)

Next week: Part 2: How proper dashboard design techniques extend to all software design (and in particular to web-based portals)