Top Retail Websites Not Getting Faster: Average Web Page Load Time Is 7.25 Seconds [Report]
According to a recent report published by Radware, load times for the top U.S. retail websites are 22 percent slower than in December 2011, continuing to go well beyond the ideal three-second load time. For their Spring 2013 “State of the Union” report, Radware evaluated load times for the top 2,000 U.S. retail websites as […]
According to a recent report published by Radware, load times for the top U.S. retail websites are 22 percent slower than in December 2011, continuing to go well beyond the ideal three-second load time.
For their Spring 2013 “State of the Union” report, Radware evaluated load times for the top 2,000 U.S. retail websites as ranked by Alexa.com. The sites were tested using three different browsers: IE 9, Firefox 17 and Chrome 23, all on a DSL connection.
The study found that the median load time for first-time visitors to a retail website’s home page was 7.25 seconds, a 22 percent increase from the load time of 5.94 seconds recorded in December 2011.
With the industry’s ideal load time at three-seconds or less, anything over the ideal load time could affect a number of online metrics, including bounce rates and return visits. (Radware based the three-second ideal load time on a 2010 PhoCusWright/Akamai report.)
Radware blames the slow load times on retail sites becoming larger and more complex with an increased number of resource requests (such as images, HTML, and CSS/JavaScript files).
In December of 2011, the median page contained 73 requests. That number jumped 8.22 percent in December 2012, with the median page containing 79 requests. Each resource request can take anywhere from 20 to 50 milliseconds for the file to make the round trip from the user’s browser to the host server and back again to the user’s browser.
Another key finding from the report showed that the top 100 retail websites are under-performing when compared to all 2,000 retail websites. The median top 100 website’s had load times of 8.23 seconds, while the overall median load time was 7.25 seconds. The top 100 retail sites are also slowing down at a faster rate than the entire list, with load times increasing 28 percent in the last year compared to the 22 percent overall.
Firefox Outperforms IE & Chrome
While a similar study of retail website load times conducted last September by Strangeloop found IE 10 to load faster than Firefox or Chrome, this month’s Radware report discovered that Firefox 17 experienced the fastest median load time at 6.64 seconds, an 8.4 percent faster load time over IE 9’s 7.25 seconds. Chrome 23 also outperformed IE 9 with a median load time of 7.09 seconds.
What Are Retail Sites Doing Wrong?
Radware’s report identified a number of ways retail websites could speed-up their load times. Only 25 percent of the 2,000 websites evaluated used a content delivery network (CDN), a service that allows site owners to cache static page resources in geographically distributed servers. Using a CDN could shorten server round-trip times, reducing load times by up to 30 percent.
Radware also recommended site owners enable keep-alives to control the number of times a TCP connection takes place. Too many TCP connections, the process by which a user and server send and receive acknowledgement that a connection has been made and data can be transferred, can slow down a site.
More than 20 percent of the 2,000 retail websites evaluated failed to compress page resources, increasing the number of bytes sent over the network and further slowing down web page load times. By Radware’s calculations, implementing keep-alives and compressing page resources could improve load times by up to 31 percent.
As websites grow in size and become more complex, this slow load time trend will only get worse. “This is a massive drop in performance,” said Joshua Bixby, Vice President Application Acceleration at Radware, “If this slowdown rate goes unchecked, we will see median load times of nine-seconds or more, which is simply unacceptable for online shoppers.”
Opinions expressed in this article are those of the guest author and not necessarily MarTech. Staff authors are listed here.
About The Author
