A Design Problem

A reader sent me the following display, which was developed by a fellow named Paul Williams to provide companies a way to determine the best of multiple options (ideas A, B, and C in the example below) across differently weighted variables (ease of implementation, etc.).

[Scroll down to see our solution to this display's design problems.]

Poor Design Example

The most useful part of this display is the small “score” column in the bottom-right corner. The formulas displayed to the left of the score column are supplemental, at best, and probably completely unnecessary.  The most salient part of the display, the blue, orange, and green vertically oriented lines that cross the four numbered axes are misleading. This is partly because the four variables that they display have different weights, but these different weights are not reflected graphically. The other major problem is the fact that by using a line to connect the four values for each idea, the display draws attention to patterns formed by the lines and their intersections that aren’t particularly meaningful. What our attention should be drawn to are the differences in magnitudes in scores among the three options for each variable and for their overall performance. The following is a better design:

Design Example Solution

When lines are used to display quantitative data, people focus on the pattern that they form, but with discrete variables such as these, which can be arranged in any order, the patterns formed by lines are meaningless. Bars help people focus primarily on the magnitudes of individual values and how they compare to one another, which is the right way to read this information.

As you can see, the lengths of the quantitative axes on these graphs have been scaled to reflect their differently weighted values. As such, people can meaningfully compare the lengths of the bars in each of the graphs to see their relative scores. I have encoded the Overall scores (the most important information) graphically and made them a darker color so that they stand out as different from and more important than the individual variables. Unlike the original display, which only has one small section of three numbers that are useful and not misleading, all of the content in this redesign is meaningful and helps people see how the variables relate to the final conclusion.