Websites for the top 500 US retailers are still too slow and continue to get slower, according to State of the Union: Ecommerce Page Speed & Web Performance, Fall 2013, a new study by Radware.

Pages have slowed down 14% since Summer 2013, and are 16% slower than Fall 2012.

Additionally, the report uncovers inconsistent adoption of core best practices by site owners, critically affecting website performance and customer experience.

Other key findings include:

- The trend toward bigger, slower pages continues. The median page took 8.56 seconds to load for first-time visitors, representing a 14% slowdown over the median of 7.48 seconds recorded three months ago (Summer 2013).

- The median page takes 5.3 seconds to become interactive. Sites have experienced a slowdown of 8% since Summer 2013, when the median time to interact (TTI) was 4.9 seconds. Ideally, pages should be interactive in 3 seconds or less.

- Three common design practices are failing users. Most sites made at least one of three critical mistakes in the design and presentation of their feature content: loading feature banners last; placing a call-to-action at the bottom of feature banners; and/or not implementing a call-to-action at all.

- The adoption of performance best practices is inconsistent, even among leading sites. Among the top 100 sites, adoption of some best practices is nearing the saturation point, whereas others remain neglected.

- Browser vendors are not keeping pace with page demands. Across all three major browsers, performance is trending downward as browser vendors struggle to keep pace with the demands of today’s large, complex, dynamic web pages.

TTI is the point at which a page displays its primary interactive content (e.g., feature banners with functional call-to-action buttons). Sites have experienced a slowdown of 8% since Summer 2013, only 18% of the top 100 sites had a TTI of 3 seconds or less, while 26% of sites took eight or more seconds to become interactive.

“In just the last few years, web page speed has migrated from the fringe to center stage, emerging as not just a technology trend, but a hot-button business issue,” said Tammy Everts, web performance evangelist, Radware. “Numerous studies have found an irrefutable connection between load times and key performance indicators, such as conversion rates and revenue increases. Site owners need to understand that optimizing performance is much more nuanced than just pushing out faster pages to customers: it’s about understanding what users want from every page of your site, then fine-tuning those pages to ensure that critical content loads first instead of last. A site owner who neglects core performance best practices is missing out on significant opportunities to make relatively easy performance gains.”

Another recent APMdigest blog: Cyber Monday Warning: Ecommerce Sites Too Slow

About the Study: Radware’s quarterly “State of the Union” report measures and tracks the performance and page composition of the top 500 US retail websites (as ranked by analytics firm Alexa.com) over a two-day period with the purpose of gaining ongoing visibility into the real-world performance of leading Ecommerce sites.

Website load tests were conducted between August 29-30, 2013, using Internet Explorer 10, Firefox 22 and Chrome 29 on a DSL connection. The tests in this study were performed using WebPagetest.org, an open source project primarily developed by Google, which simulates page load times from a real user’s perspective. Radware tested the home page nine times of every site listed in the Alexa Retail 500.

In addition to measuring a core set of metrics – load time, resource requests, page size, implementation of core performance best practices – the study also measured each site’s median time to interact (TTI). Time to interact is considered to be a more meaningful indicator of a page’s ability to deliver a satisfactory user experience to a visitor, providing additional insight into real-user performance. To identify the TTI for each page, Radware generated timed filmstrip views of the page load for the median page for each site in the Alexa Retail 100.