Speed of a Site and Usability

September 24th, 2007  |  Published in Design, Usability, Web Development  |  1 Comment

Jared Spool and Christine Perfetti discuss a study on how web page speed impacts usability on their latest Usability Tools Podcast. Because this study conflicts with some of the research that I cited during my recent presentation to DevGroup NW on ways to speed up your site, I was anxious to listen to the podcast and review the research.

Basically, the UIE study found that speed did not have the impact on usability that everyone in the human factors field believed. This conflicts with the research cited in Andy King’s Speed Up Your Site book which found that the speed of systems had a high impact on usability.

Here is a quick summary of the findings:

  • The major finding is that the a strong correlation between perceived download time and whether users successfully completed their tasks on a site. In other words, if someone completes their task on a site successfully, they will feel the site was faster than it truly was. If they can’t succeed, they will perceive it as being slower.
  • The secondary finding was that the speed of the page had no correlation with whether or not someone would could complete their task successfully. So making a site faster doesn’t necessarily make it easier to use.
  • In the podcast, Jared described a Gesalt theory from the 40s that time is perceived as going more slowly when your in pain and more quickly when things are pleasurable. Frustrating websites have the same impact on perception. When the website is enjoyable, you perceive it to be faster.

At some point, I’d like to examine the UIE results and try to reconcile them with the other research that has been done in this area, but given the respect I have for Jared Spool’s work, I’m going to accept the UIE study as definitive.

Here are my thoughts on the study and what it means for those looking to speed up their sites:

  • Perception matters more than reality. This was one of my main points of emphasis during the presentation. During the presentation, I focused on how you can build the page to make it seem to load faster. The UIE study says that having a usable site has a big impact on speed perception so looking at the ease of use of the site should be a top priority (if it wasn’t already).
  • The study focuses on web pages, not applications. As far as I can tell from the study and the podcast, the study focused on web pages and sites, not web-based applications. I believe one of the reasons that Yahoo has spent so much time focusing on speed is because of they want people to do work—repetitive tasks—in their applications. I propose that people have less tolerance for delays in their applications than they will for web sites they visit.
  • Other studies still show speed impacting credibility and shopping cart abandonment rates I can’t believe all of the research on these topics is inaccurate. The latest was from Jupiter Research showing 4 second download thresholds for ecommerce sites.
  • The cost-savings for speeding up your site are still worth it. During the podcast, Jared acknowledged that speeding up a site can save money on bandwidth.
  • You shouldn’t have to choose between speed and usability. This is the place where I felt the conclusions of the conclusions of the study were off-base. The study says, “what we’re seeing leads us to wonder if it’s worth the resources to make web pages load like lightning.” And during the podcast, Jared gave the example of companies spending hundreds of thousands of dollars to optimize their websites for speed while ignoring core usability issues.

    Speeding up your site doesn’t require hundreds of thousands of dollars or lots of resources. That was the main point of my presentation last week. Some of the things that speed up sites the most are brain-dead simple (e.g., turn on gzip and shrink page sizes 70 to 80%). This shouldn’t be an either you spend the money optimizing the site or you send the money on usability testing question.

Those organizations that are spending hundreds of thousands of dollars on speeding up their sites are probably looking in the wrong spots for the speed improvements. They are probably spending their time with expensive efforts to increase their server and database speed while ignoring the reality of Yahoo’s 80/20 Rule that the majority of the savings come from frontend design decisions.

So yes, web sites should be designed to be usable and have utility. These efforts are crucial and deserve priority. But very simple changes—mainly gzip and reducing the number of http requests—can have a huge impact on speed. There is no reason to chose between usability and speed.

Responses

  1. Jared M. Spool says:

    October 1st, 2007 at 10:03 pm (#)

    Your conclusions are right.

    One detail: the Jupiter study decided 4 seconds was a threshold by asking people how fast the site should be.

    We asked users about the sites. But, in our study we also watched the users interact with sites and timed the pages. What users said was fast actually wasn’t that fast. You can’t always rely on what users say. (In fact, you can only rarely do so.)

    I wonder what conclusions the Jupiter folks would’ve come to if they’d actually watched the users interact with the sites and timed the pages.

    Thanks for encouraging our behavior,

    Jared

    p.s. When you have a site with 10,000,000 (or more) pages like our clients do, you’ll find that many things cost hundreds of thousands of dollars. Even if it only costs $0.10 per page.