| About InfoWorld : Advertise : Subscribe : Contact Us : Awards : Events : Store |
 |
||||
|
||||
 |
||||
 |
|
|||
 |
||||
|
|
|
||
|
|||||||||||||||||||||||||||||||||||||||||
 COLUMN
The speed of business: If your pages are slow, your customers will go SPEED KILLS, BUT not on the Web. When it comes to the Web, we love life in the fast lane. The faster your pages download, the more likely your customers are to stick around.
Slow pages translate into unimpressive revenues. Zona Research's study, "The Economic Impacts of Unacceptable Web Site Download Speeds" (www.zonaresearch.com/deliverables/white_papers/ wp17/index.htm), concludes "that at a run rate of about $362 million per month, perhaps as much as $4.35 billion in e-commerce sales in the U.S. may be lost each year due to unacceptable download speeds and resulting user bailout behaviors." If your pages are slow, you're inviting your customers to visit the competition. There's a de facto rule that page loads should be no slower than six to eight seconds. That rule is too forgiving -- sites should deliver pages no slower than 4 seconds. In the 1980s, IBM conducted a study that determined successful response rates for what were then terminal-and host-based applications. It concluded that response rates should be no more than 2 seconds. Beyond that, productivity declined significantly. I understand the difference between Web-and host-based systems, but the difference between 2 and 8 seconds is far too much. Our expectations appear to be heading in the wrong direction. I'm also certain we didn't become more patient over the years. Can pages load in 4 seconds or less? Is this achievable? You bet. Check out Keynote's Business 40 Internet Performance Index at www.keynote.com/company/overview/public_services/business/ bus40_index.html. Sites such as Yahoo, Schwab, and Microsoft average far less than 2 seconds. The Business 40 average is far less than 4 seconds. So, is your site slow? Maybe you've heard from your customers, but the majority won't take the time to tell you your site is slow. They'll simply leave. You could and should measure your page loads yourself, but be prepared for less than accurate and extendable results. Go with a service such as Keynote, although there are other tools available from companies such as WebTrends, Mercury Interactive, and more. I favor the Keynote outsourced subscription approach because the service is both hands-off and distributed. I also favor the distributed model where computers measuring performance are strategically placed throughout the country. Lots of folks cheat themselves when it comes to performance tests. For example, many don't time end-to-end response rates. To do your business and your customer justice, time your page loads from when the customer hits the key or clicks the mouse to load your page until the page is completely loaded. If your pages connect to other sites such as advertising servers, be sure to include these times in your count. Partial metrics are valuable when it comes to optimizing or troubleshooting performance bottlenecks. Once you know your end-to-end response rate, you can drill down and see what's working and what's not. I asked the InfoWorld.com team what made for fast pages, and I received a diverse list, including such items as load balancing, caching content, server and network capacity, page size, and efficient page coding. In upcoming columns, I'll be breaking down speed-related issues. Before then, determine if your site could use a speed lift. Laura Wonnacott is vice president of InfoWorld.com.  RELATED SUBJECTS  E-CommerceWeb TechnologiesMORE > SPONSORED WHITE PAPERS 
SPONSORED LINKS
|
|||||||||||||||||||||||||||||||||||||||||
|
||||||||||
|
||||||||||
