Posting Date: September 10, 2003

How Critical is Page Download Time by Itself?

Abstract: The total time of user activity on every other web page and every other task is the real reason why users are only willing to wait 5-10 seconds for a single page to load. Users are actually quite patient regarding single, one-time page downloads.

The time it takes for a user to download a web page is a very small part of the overall time it takes for a user to get what they want. We can probably assume that it takes several seconds for a user to decide that they want something, such as information, news, or a product. It might take a lot longer than that.

Next, it will take the user several seconds to get to a point where they can even begin their actual quest. For example, the user might need to boot their computer, log in, and launch a web browser. This could take a few seconds or many minutes. At this point, the user still isn't concerned with download time but the overall user experience clock is ticking.

Once the user finally opens a browser, there is the decision (i.e., time) to go directly to the web site (e.g., Amazon) or to do some sort of search (e.g., Google). This also takes time. Now, once a user has made it to the destination web site more work needs to be done. Searching, navigating, wondering, looking, thinking. This takes time too, of course. Only once the user actually is ready to get to the destination web page can we talk about page download time. The activity accumulates before the destination web page load time. Too many "experts" completely ignore this first investment in time. And yes, this really is an investment for the user, because it is a committment. It is also a risk for the user, with the hope of a return.

Most folks concerned with page loading time are actually talking about the time it takes for a destination web page to load. However, for the user, the clock will definitely still be ticking even after the destination web page loads. For example, if the destination page is a news page, the user still needs to read the content. Or, if the user intends to buy a book, there is still the time it takes to scan the page, perhaps read the reviews and summary, and find the button to add the item to the cart. If the user is buying something, like a book, it will take time to checkout.

We are still not done talking about time because if the user needs to wait for delivery the user is still concerned with time. The moral of the story is that even if a destination page loaded in less than one second (wow!) the user still measures the web site in terms of minutes, days, or even weeks. Remember, the web site is the company. And while it might be nice for web developers and web usability folks to think they only need to care about page load times, or even the web site itself, they are completely wrong. Time for users is measured by their entire experience, not something as narrow as page load time for a destination page.

I call into question research that is concerned with single page (i.e., destination page) download times. Why? Because I would argue that almost any user would be willing to wait between 5 and 10 times as long for a single page to load if they were highly confident about it. If success is almost certain, they'll place a big bet on it (i.e., wait time). They will wait a long time, and when it does load, they'll be happy. As Christine Perfetti indicated, success drives the perception of page load time. Not surprising, especially in light of total task time as I am suggesting.

Here's an example. Imagine one task that was designed where the user only needed to rate satisfaction (for example) of page download time but did not need to wait for any other pages to load, and did not need to do any work before that destinatation page loaded. Imagine another task where the user needed to do work to get to the page (search and navigation), but then had to wait for the exact same destination page. Which task would yield higher satisfaction ratings? The one where the user only needed to load the destination page, of course!

In spite all of the articles I have read, I think many users are quite patient. But for them we need to remember that the clock has been ticking for a long time before they reach their destination pages. Furthermore, the clock is often ticking long past the time they were on the destination page.

Reader Comments...

One more quick comment. Many people forget about the importance of the "weakest link" of the entire experience. Anything could be the weakest link. More important is the fact that, as the user does more and more, the opportunties for failure compound significantly. If the user only needs to visit 2-3 pages to get to the destination page, there is a good chance of success and the weakest link will not be broken. However, if the user needs to visit 9-10 pages (or perform a large number of actions) the weakest link will be more of a threat. The user will become more and more frustrated over time, and the weakest link is more likely to break. If that weakest link breaks, no amount of time can salvage the overall experience. So, we are doing battle not only against time but against all of the other possible failure points on the way to the destination web page. Yikes!

Posted by: John S. Rhodes on September 10, 2003 10:11 PM

I take the points about other parts of the user's investment of time -connecting to the ISP, loading the browser, etc.

But I still think download time is important, for example:
* Few sites have no competition. Most have to compete, and one aspect of that is having at least reasonable load times.
* The user might not want to visit furhter pages on a site where the first page is much slower than the average in that user's experience of the web.

Posted by: Philip Chalmers on September 11, 2003 07:34 AM

My memory is failing me — there's an Australian site with a maritime theme to it, deals with internet issues. Guy's name is David something... He wrote about a lot of this a couple years ago, and used an economics metaphor (much like the one used here), referring to users as in an options market... It was valuable to me, because we had heard focus group respondents say that they have click-hesitancy based on prior bad experiences with download time on the site. I tried using that as an argument that the Web site was a community, that every page manager helped every other page manager by improving their pages. Damn damn damn, wish I could remember the site! AGH!

Posted by: Frank on September 11, 2003 09:00 AM

I agree with what you say, and would even back up the experience before the computer boot to the point where the user says/thinks, "I need a ____ and only the Web can provide it."

However, I would bet most of "us" are not in control of the entire user experience. We control only a few of the pages. As such, we should be aware of page loading times. Not as a primary concern or course. I still get people at work using 54 lines of code and 5 images (logo and 4 byte-heavy spacers) to make a page header instead of 34 lines (I could get fewer with more than the 10 minutes I spent), 1 image (logo) and a CSS file.

From a Web app perspective, I look at total task time compared to all the other tasks the user has to accomplish, broken out as appropriate to the task (hourly, daily, weekly, quarterly).

I think that would be another differentiating point: Web site vs. Web app. Information foraging vs. transaction completion. My guess (definite guess) is that people will be less patient with the perceived performance of a Web app, especially once they get used to how it works. Initially they may be more forgiving, because they will blame themselves for "mistakes."

Here's some other info (via usability.gov) on page load timings.

Posted by: Matthew Oliphant (formerly fajalar) on September 11, 2003 09:27 AM

It's all about setting expectations. If you don't tell people what they are waiting for (giving them a reason to wait) they won't. Or, if they will, they'll be more likely to be dissapointed if they waited and the content wasn't worth the wait.

Posted by: Jason Fried on September 11, 2003 09:27 AM

My wife is probably a typical user. When a site loads slowly, she will wait long enough to be convinced that something has gone wrong with the connection, then go to the address bar and hit Enter to try to reload the page. She gets annoyed when this happens, but at "The Internet," rather than the site that won't load. She is usually very persistent. She never looks at the little progress bar at the bottom of Internet Explorer (which takes an experienced user to interpret correctly anyway.)

Posted by: on September 11, 2003 09:45 AM

they'll be more likely to be dissapointed if they waited and the content wasn't worth the wait.

Overall I agree, but I do think there is a point, based on past experience with similar (or the same) content/transactions where the user switches from blaming themselves to blaming the Web site.

Not that I am anywhere near a typical user (if there is such a thing), but I still find I blame myself for somethings I do on the Web, when in retrospect of the action I realize that it was the site's/designer's fault.

The user should never feel bad about making a decision while using one of our products or services. It's not as detrimental to the business if they are in the "I blame myself" stage, but it becomes detrimental when they enter the "I blame you" stage.

*sigh* Being sick at home leaves far too much time to write posts. :)

Posted by: Matthew Oliphant (formerly fajalar) on September 11, 2003 09:52 AM

Download times are a barrier to adoption. 56k is the ideal most never achieve because conditions are never perfect. I have two relatives who I got online with 6-months-free ISP offers. After six months they had abandoned the web and dropped service. Why? Web sites are too damn slow to load. Designers have smoked fat pipes too long and can't remember the reality that most of us occupy. There is a new caste system rising -- wired folk who can afford broadband, analog cretins who eak by using e-mail mostly because the web is a nightmare, and the unwashed masses who can't afford to spend ten hours a month waiting on their status bar.

Posted by: Moe Canby on September 11, 2003 11:00 AM

John,

As usual you are provocative in your posts, which evokes great discussions (perhaps your intent). I disagree of course. Your statement that users are "quite patient" when waiting for pages to download is flawed I think. I'd like to see studies that show this effect. I do agree that the total experience is important, and there is some evidence that "cumulative frustration" may have some influence on user satisfaction.

However, the research I summarized in my book, Speed Up Your Site, shows that page download times are important to user satisfaction. The UIE "study" above is often cited, but they haven't revealed their data. Every other study I've read contradicts what you propose. Users hate to wait and there is usually an alternative site that they can go to. Speed matters, and it is a critical part of usability, utility, and likability (Shackel).

I've seen sites that were 1.5MB and fatter for their home page alone. This is an abuse of the user's time, if they ever wait around for the entire page to load. Slow pages say in effect, "we don't value your time." They are user hostile, not user-friendly. Fat graphics and Flash splash screens are out. Fast and sleek and standards-compliant are in.

Posted by: Andy King on September 11, 2003 11:55 AM

Related comment:

One thing I think we cannot trust is any "usability study" on this topic. Contextual Inquiry reports (low interaction), observation sessions (low interaction) in the user's environment, and possibly interpreting some clicking data. But any "study" is going to be with user's who "want to help" and they will wait as long as it takes (in many cases) for the page to load. And even when the pages take forever to load, they will still say they had a satisfying experience. If you have a good facilitator that can delve without pushing, you may get some decent data, but it's doubtful.

Andy asked for some studies to back up John's claim. They exist. Not at work today, otherwise it would be easy to point to them. But there are some sources in the link I posted to usability.gov above.

Lastly, I highly approve of using quote marks around the word "study" when used in the same sentence as UIE.

Posted by: Matthew Oliphant (formerly fajalar) on September 11, 2003 12:10 PM

As mentionend before, expectations are an important factor in this game - and the importance of the task itself. If I expect to find the answer of my request at a special page I will wait - even for a long, long time. My patience correlates directly with the importance of the activity. But if I step through some search result pages looking for some less important results, my expectations at most depending on the description and title of the listed pages, the download time is a major factor.
An additional note: Imho it would have been better to compare sites with relating themes/content and not Amazon with About.

Posted by: Gerald on September 11, 2003 08:58 PM

Give me a device I can switch on in the morning and *bang* its ready to go. No splash screens, no need to get a coffee while waiting ... Why cant switching on a PC be as fast as turning it off? Then again, is too fast, too fast? I mean do people require a certain amount of waiting time for certain tasks? I could always just leave the PC on :)

Posted by: Daniel Szuc on September 11, 2003 09:30 PM

One of the good things about the old 28.8 (went from 28.8 to DSL) days was I had to get up and make food from time to time. Now with DSL at home T3 at work, there's just no time anymore to enjoy a good walk to the kitchen, or even talk to the people who live with me.

So maybe waiting is a good thing.

Posted by: Matthew Oliphant (formerly fajalar) on September 11, 2003 10:17 PM

By choosing to focus on computers rather than the tasks we wanted done, we inherited much of the baggage that had accumulated around earlier generations of computers. It is more a matter of style and operating systems that need elaborate user interfaces to support huge application programs. These structures demand ever larger memories and complex peripherals. It's as if we had asked for a bit of part-time help and were given a bureaucracy.

Canon Cat

Posted by: Mac on September 12, 2003 05:22 AM

Another thought, think it also depends on what you are looking for, is it worth the wait and if there is competition who can provide something faster. For example, Google provides me with everything I generally need from a Search perspective. But if there is something out there that can offer something faster and better, I would probably use it. I also wonder if there have been time comparisons done on finding information or completing a task using the web vs more traditional channels. For example, if I can do it faster using the phone, why would I use the web site to do it in the first place?

Posted by: daniel szuc on September 12, 2003 06:35 AM

I also wonder if there have been time comparisons done on finding information or completing a task using the web vs more traditional channels.

Yes, there have. In studies that set paper vs. electronic for information search and comprehension, paper always wins. Of course most of the people tested have YEARS of experience looking for things on paper medium, and electronic is newer. I wonder if there are any trending studies.

If you have access to HCIRN (I don't) there are some related papers.

Posted by: Matthew Oliphant (formerly fajalar) on September 12, 2003 09:10 AM

Posted by: Mauro Gutkii on September 12, 2003 02:38 PM

Mauro, you're correct... I was able to recall the url some time after posting but I still couldn't find the specific column I remembered, even using the search engine. I emailed him asking for assistance, but haven't heard back.

Posted by: Frank on September 12, 2003 04:29 PM

Subscribe to Webword.com
Receive the best free usability newsletter on the Internet.

URL: http://webword.com/weblog/