The Importance of Site Performance

Site Performance: The Need for Speed

In a 2006 study on site abandonment (.pdf) conducted by Jupiter Research on behalf of Akamai, 28% of online shoppers claimed they would not wait longer than 4 seconds for a page to load before leaving a site. This equated to 33% of broadband customers and 19% of dialup. Though broadband customers had higher expectations for page load speed, even 45% of dialup users were unwilling to wait more than 6 seconds for a page to load.

Online power shoppers who spend more than $1500 online are more likely to demand fast loading pages (55% vs. 40% of shoppers who spend under $1500 per year).

Speed’s Impact on Loyalty – Online and Offline

Faster sites attract more repeat visitors and customers. 64% of dissatisfied online shoppers said they were less likely to visit a slow retailer again, and 62% were less likely to purchase from the site again. 48% would purchase from a competitor, 28% would hold a negative perception of the company, 27% would tell a friend about the bad experience and 16% reported they would be less likely to visit a retailer’s offline store after a bad online experience.

For all these reasons, Jupiter and Akamai concluded you should shoot for a 4 second or less page load.

It’s 2009 – What’s Changed?

Akamai teamed up with Forrester Research earlier this year for a follow up study and found online shoppers have even higher expectation for web performance — and poor performance has an even bigger impact on customer loyalty. For example, nearly half of consumers do not want to wait longer than 2 seconds for a page to load.

How Do Online Retailers Currently Stack Up?

In 2008, Internet Retailer conducted a retailer survey and found:

  • 68% of retailers’ home pages load in under 3 seconds
  • 43% load in less than 2 seconds
  • 81.8% load in 30 seconds or less for dialup users, with 50.6% under 15 seconds
  • 43.3% say the use of video, animation and AJAX has hurt site performance

Do Online Retailers Monitor Performance?

  • 77.3% monitor site responsiveness
  • 9.1% engage in load balancing testing and content validity
  • 4.5% measure application behavior

What Do Online Retailers Measure?

  • 73% measure site performance’s impact in terms of lost revenue
  • 59% measure lost traffic
  • 43% measure the increase in call center and email traffic
  • 27% measure the increase in negative customer reviews
  • 21% measure the impact on customer satisfaction ratings
  • 40% benchmark their performance against competitors

What Do Retailers Test?

  • 37% test home and product pages consistently using real-time reporting tools (26% test daily, 19% test weekly or monthly)
  • 84% test various screen resolutions
  • 83% test across different browsers and operating systems
  • 16% test performance from different geographic locations or different times of day
  • 16% record and replay specific user transactions
  • 13% test performance of mobile applications
  • 53% conduct testing before the holiday season

The Internet Retailer article also suggest retailers retailers should test the performance of new page treatments (including A/B testing) and custom applications they add to their ecommerce platform or web hosting service before full roll out. It reminds us that the features and functions may work fine in the environment in which they were built (design and development firms having top of the line systems), but differently when installed on the retailer’s platform.


Related Articles

16 Responses to “The Importance of Site Performance”

  1. One of the arguments made against eBay is that Item Description pages load too slowly. In recent years, the trend to slower page loads has increased to a frustrating degree.

  2. Hi Linda, just to clarify: This post in a couple of different places notes that the webinar is “tomorrow”, but also gives 11/4/2009 (Wednesday) as the date of the webinar. (“Tomorrow” would be Tuesday 11/3, as today is Monday 11/2.) Which is correct? Thanks!

  3. Andy says:

    I’d be interested in hearing what tools everyone is using to measure page loading speed.

    • @Andy, there are tools like LoadRunner, JMeter, WebLoad or 3rd party monitoring/optimization services like Akamai and Gomez. Yes, would be interesting to hear what most online retailers use, and what Get Elastic readers use.

  4. Interesting post. I think an often overlooked part of site performance is the “perceived speed” that the visitor experiences. E.g. a site which loads in 6 seconds might be perceived fast by the visitors if all the slow loading elements is below the fold and the top of the site loads in less than 1 second. On the other hand a site that only takes 2 seconds to load, but where 1,95 seconds is the graphics and styling of the navigation menu might be perceived slow.
    My point: it isn’t the actual page load time that matters, it is the perceived load time thats important.

    @Andy; if you just want to measure it manually yourself install the Firebug add-in for Firefox or use the “developer -> webinfo -> resources” function in Safari.

    • @Christian, yes this is also important and should be measured at the browser level too. I saw some research by Gomez that showed the actual load in IE and Firefox vs. perceived speed. IE was 5.335 s but perceived as 1.614. Firefox was actual 4.08 and perceived as 2.461! All depends on the browser, which objects it decides to load first, and whether they are the most important objects on the page.

  5. Looking forward to the webinar. Page load times definitely have an impact on the bottom line. We had a recent blog post about this data as well:

    Continuous testing of changes as they relate to page load times can make a difference in converting more sales.

    Rob – LexiConn

  6. JakeFras says:

    The corporation I work for has 3 large ecommerce sites that sell internationally. It took me 12 months to get funding for web performance monitoring software but then my employers were finally able to grasp that our customers around the world were experiencing different page load times for our catalogue pages and this in turn was effecting abandonment rates and consequently sales.

    I recently read that when Shopzilla increased the speed of their website by 5 seconds it resulted in a 25% increase in page views, a 10% increase in revenue, a 50% reduction in hardware,
    and a 120% increase traffic from Google – that’s crazy!

    How many of you use Adwords for leads? We do, we spend 10′s of thousands on it. Did you know that landing page load time is an Adwords quality score factor? The slower your ad landing page loads the higher the ad rate you pay. Web performance is important.

    Early on in the year Steve Souders, Google’s Web Performance guru mentioned a website accelerator by a company called Aptimize – this is what we implimented and it has made a significant difference to the page load speeds of our sites. Web performance is now taken very seriously where I work especially coming up to a holiday season. Great article and good to see this issue is being aired.

    • @JakeFras, yes site performance is the hidden jewel in ecommerce optimization – hidden because we’re usually looking at features/design elements when optimizing landing pages…

  7. Thank you for your sharing,I like this.

  8. @Linda, yes my point exactly. Do you have a link for the Gomez study? Was it in one of their case studies found here: ?

  9. JakeFras says:

    @Linda – agreed, the big sites e.g. Shopzilla, Amazon, Google, Yahoo etc get it; they have teams of people dedicated to shaving even miliseconds off page load time, yet the vast majority of us have no idea that our webpages appear at different speeds for different visitors.

    I guess its not surprising that this ‘problem’ is largely unknown because most of us assume that people’s speed experience of our web pages is the same as our own – if our site is fast for me its fast for everyone, right? Wrong. In reality, the further the distance between where a website is hosted and the visitor, the longer it takes for a page to load. Web performance is all about reducing that page load time over distance using various techniques.

    To be honest I knew zero about web performance until about 2 years ago but I get it now – and I am a business analyst and not a developer or engineer. If you’re looking to ‘dip your toe’ in this subject the Akamai webinar would be a good place to start.

  10. Drew says:

    @JakeFras – Aptimize looks really cool. Too bad I don’t see much support for the java stack, but i didn’t read to far into the website.

    A poor person’s version that I’ve run into in the java world is JAWR (, which keeps in mind the need for keeping UI optimized files maintainable for the devs/designers.

Leave a Reply

© 2014 Get Elastic Ecommerce Blog. All rights reserved. Site Admin · Entries RSS · Comments RSS