Shoppers are not patient.
Consider yourself. How often have you gone into a brick-and-mortar store, had a quick question about a product, but had to wait for an answer because there was no clerk to be found?
Most shoppers have experienced what could be called bad service at a local store, and no one likes it. Unfortunately, online stores can also keep shoppers waiting. We are talking about a few seconds compared to several minutes, but many ecommerce sites simply take too long to load.
To confirm this point, I took a survey of 25 online retail websites. In this non-scientific survey, I visited each domain twice, ran a test with Yahoo's YSlow site performance measurement tool, and recorded the better of the two test results. In some cases there were significant changes in performance from one test to the next.
YSlow Tests for Best Practices
I've had the YSlow Firebug plug-in for a few years, but after interviewing JavaScript performance expert, Nicholas C. Zakas, I have been much more concerned with performance, and I have a newfound respect for YSlow.
The tool measures up to 35 known site performance best practices, offering both an individual grade for each best practice and an overall grade for the site.
Survey Results
Of the 25 sites I surveyed, Williams-Sonoma's website earned the highest grade, a B. The site was generally A material expect for a few key components, not the least of which were HTTP requests.
Estimates from leading site performance experts like Steve Souders and Tenni Theurer suggest that 80-to-90 percent of page load times are devoted to HTTP requests. Simply reorganizing external JavaScript files, style sheets, and images can significantly improve site performance.
Other sites did not do as well as Williams-Sonoma. Amazon, perhaps the best-known online retailer, scored a C in YSlow with an overall grade of 79. Amazon was actually the reason that I took more than one test sample, since in the first pass the site scored a low D. I also discovered that Amazon's score changed based on whether or not I was logged in. In the better of the tests, Amazon loaded six external JavaScript files, four external style sheets, and seven background images for a total of 17 HTTP Requests.
Barnes & Noble, which is my favorite online bookstore, was graded an E (and yes there is an E grade) for making 91 HTTP requests, the most of any of the sites surveyed.
