We’ve seen several examples of how synthetic tests report data. Let’s look at a few other common representations of end user experience. One way of reporting performance is to segment it by region or carrier to identify the worst-served parts of the Internet, as shown in Figure 9-28.

Figure 9-28. Webmetrics transaction performance report by geographic region for a site on the East Coast of North America

Segmented reports show you the slowest or least-available tests along a particular dimension. But what if you want to focus on a particular element of latency? Synthetic tests collect a great deal of data, and you need to look across all of it to properly understand what normal performance is like.

By displaying a metric—such as total transaction time—as a histogram, you can quickly see what kind of performance is most typical. On the other hand, you may not want to aggregate data, but rather see individual tests to better understand a particular period of time. In this case, a scatterplot visualization like the one shown in Figure 9-29 works well.

Figure 9-29. A scatterplot of individual test results

This kind of report is only available from services that keep the raw test data for considerable lengths of time. Some lower-cost services will aggregate this information into averages to reduce their storage requirements.

Finally, matrix reports are ways of comparing two segments to one another. In the case of Figure 9-30, these segments are locations and websites.

Figure 9-30. Gomez matrix report of websites by region

Matrix reports are useful for identifying problem areas at a glance, such as two carriers that aren’t forwarding packets to one another well. From a matrix report you can drill into the history of a specific region or site.

As you define and prepare these reports, be sure to share them with managers and members of other monitoring groups, including marketing, QA, and web design, to accustom them to the information that synthetic monitoring can provide.