Using WebPageTest

Chapter 5. Simulating Real Users

In Part I, the first thing we said about WebPageTest was that it is a synthetic tool, not to be confused with real-user monitoring. By virtue of being synthetic, WebPageTest is only an approximation of page loading performance. But synthetic testing can (and should!) still accurately resemble the way actual users access the page.

Why would it matter if a synthetic test is realistic? It is completely within the realm of possibility that a user with an empty browser cache from Dulles, Virginia in Internet Explorer 9 over a cable Internet connection is visiting a given web page. The goal of synthetic testing, however, is not necessarily to optimize for one particular use case but rather to optimize for the entire user base. The best way to do that is to set up synthetic tests that are representative of the population. For example, test in a commonly used browser from a typical location over reasonable network speeds. These simple changes allow you to focus on the performance issues that actually matter to users.

To make the point, imagine that you’re contracted to optimize the performance of a university web page for enrolled students to check their class schedules. You can assume that users are likely to be on very fast university infrastructure, and there’s a good chance that they will be accessing the site from their mobile phones as they figure out when and where their next class is. Given this information, a good test configuration would make ...