A clever variant of the bivariate scatter plot, the period-share chart, shows how competitor market shares change when measured at two periods in time, typically on a year-over-year basis. I have seen this technique used only rarely, but its properties make it a natural exhibit format for analyzing security data.11
To create the chart, the analyst plots each competitor's market shares relative to the leader for both periods, with the previous period on the x-axis and the current period on the y. A position above the diagonal designates the competitor as one who gained share relative to the previous period; below the diagonal, as a share loser. Figure 6-19 shows an exhibit from a Boston Consulting Group study of capital markets.12
Figure 6-19 Sample Period-Share Chart (BCG Investment Banking Study)
Copyright © The Boston Consulting Group, Inc. 2004. All rights reserved.
Examine the chart carefully; its simple appearance belies a sophisticated analytical approach. The use of ratios relative to each period's leader is the key. By using these measures as the x- and y-axis values, readers can clearly see the relative period-over-period increase (or decrease). The beauty of this approach is that firms whose positions change relative to the lead firm "automatically" appear above or below the diagonal. The period leader always falls either along the top or right of the chart. If a firm holds the leader position for both periods, it appears in the upper right, at the top of the diagonal line.
Mechanically speaking, spreadsheet software such as Excel can create period-share charts, but only after the analyst applies a little persuasion. A standard X-Y scatter plot supplies the base. To give the chart a bit of extra headroom, I recommend increasing the axis maximums a little past 100% (1.0)—for example, 110% (1.1). Period-share charts should always have an enclosing plot frame. The analyst can use one of two methods to create the diagonal line. The quick-and-dirty way involves overlaying a diagonal line over the chart. However, this means that the chart cannot be resized without also moving the diagonal line and that printed versions of the chart may suffer from line alignment problems. Therefore, I recommend the following technique:
- Create a new data series containing exactly two data points: (0,0) and (110%,110%).
- Plot the new series on the chart.
- Add a trend line for the series (this adds the diagonal).
- Change the line's thickness and color so that it matches the plot frame.
- Hide the markers for the series (leaving just the diagonal).
Shifting the focus away from general use cases for a moment, how might one use a period-share chart to represent security data? One way might be to replace the concept of "market shares" with "vulnerability shares" or "spending shares." In other words, use the period-share format to show how the distribution of security flaws or spending priorities changes over time.
Figure 6-20 shows a concrete security example. On the chart, I have plotted the "vulnerability shares" of flaws found in major security vendors' software packages for the years 2003 and 2004.
Figure 6-20 Period-Share Chart (Security Example)
Another security example might be a plot of spending priorities year-over-year, or changes in the types of threats detected by an organization's perimeter controls. I leave these as an exercise for the reader.
Period-share charts efficiently show the answer to a basic question: what's hot this year (quarter) compared to last? In that respect, they work as well as—and maybe better than—stacked bar charts or area charts. But period-share charts work less well in certain situations. For example, the period-share chart (in Figure 6-20) says nothing about the absolute number of vulnerabilities found for either the leader (Symantec) or all companies in the data set.
Therefore, as a rule of thumb, avoid period-share charts when
- You need to show absolute share values.
- You need to know the absolute size of the market.
- Data sets are thin, leading to a situation where multiple points "cluster."
- The number of competitors exceeds fifteen or more.
- The desired periods for analysis exceed two.
In many cases, alternative chart formats (such as the stacked bar chart or the Pareto chart) are a better choice.