Protecting the Golden Goose

Jared Spool

September 20th, 2005

Yesterday, I had a fascinating conversation with the marketing manager from one of the Internet’s top 100 most-visited sites. He told me how the site’s founders and the management team had decided that it was time for a redesign of their site and they wanted to know how we could help them.

This site hadn’t changed since it was launched in 1995. And because of the usefulness of the information (and, partially, the simplicity of their URL,) it had become one of the web’s most useful resources. But, as this manager claimed, it looked like a site designed in 1995. (Actually, because it had a white background, it looked to me like a site designed in 1999. A textured grey background would’ve been more 1995.) Site design has progressed monumentally since 1995 and with today’s technologies, you can do some stuff that looks really slick. I can see why they had a tremendous urge to redesign.

But here’s the problem: There is no evidence that they needed to change a thing. None. Zero. Zip.

I have heard countless stories of sites doing redesigns and having disastrous results. A few years back, one of our clients, a major big box retailer, spent more than $100 million (!!) on a site redesign — the purpose of which was purely to increase usability and, therefore, sales — only to find out, within a month after the new design’s launch that they had successfully reduced sales by 20%. It took them more than a year to make the changes necessary to get back to where they were before their “improvements”. (Think what they could’ve achieved had they only spent $50 million on their redesign.)

No evidence doesn’t mean that I think changes are bad. It just means that the team doesn’t have the information they need to know what to change and what to keep the same. And when teams don’t have the information they need, they make mistakes. Sometimes very costly ones.

In this particular case, the site is free to all users. 18 million unique visitors come to the site every month. 300 million unique page views are offered up. Advertisers pay for the site. Advertisers are paying to ensure that all those users see all those pages.

The worst thing the site owners can do is make a change that decreases their page views. A 40% reduction in page views because users don’t like the changes to the site could be financially catastrophic.

When I painted this scenario, there was silence on the other end of the phone. “Wow, I’d never thought of that,” was the next thing I heard.

Like many of the folks we talk to, these guys need to, first, learn more about their users and their customers (the advertisers):

  • Who is coming to the site?
  • Who is “paying for the site”? (Which subset of users are the advertisers most interested in?)
  • What do those users want from the site?
  • How will they react to change?
  • What changes will they see as “improvements?” (and thereby increase page views?)
  • What makes the site valuable to the advertisers? (Is it just the sheer volume of traffic? The clickthroughs they are seeing directly from the site? The quality of the visitors?)

What they may find out, when they collect this information, is that users aren’t paying any attention to the graphical look of the site. They come to get their information and they leave. Since the information is already provided in an ideal form, improvement isn’t necessary. Leaving the golden goose to lay its eggs is just fine.

If, from the information they’ve garnered, it’s clear that change is warranted, then I think they need to proceed cautiously. Make changes very locally and carefully. Only change a small portion of the site and sit back and see what happens.

Redesigns bad. Incremental change good.

With the help of technologies like CSS and the sheer number of visitors they have in a given day, it would be easy to set up measurement tools that would help them understand the impact of changes. For example, they could change key pages to be standards based, then offer up a radical style sheet to a small experimental group, say 5,000 of the day’s visitors. Watch if those 5,000 visitors behave differently than others on the site (who are still seeing the old design), possibly over several days or weeks.

Of course, field studies, usability testing, and interviews will be an important set of tools to help manage all of this. But patience and a good strategy will be just as valuable.

Have you gotten caught up in a change-for-change-sake scenario? Does this sound familiar to you? What would you have suggested to this company? Would you recommended a different strategy for redesign?

By the way, Kelly Goto is giving her viewpoint on this very topic in her UI10 session, Web ReDesign Redefined: Strategies for Success, which at the time I wrote this post, is on the verge of selling out. If you miss her at the conference, definitely pick up a copy of her book on the topic: Web ReDesign 2.0: Workflow that Works

4 Responses to “Protecting the Golden Goose”

  1. Enric Naval Says:

    I’m just going to re-write completely a site. Brand new code from the bottom-up.

    Of course, since changes in design scare people, it’s going to have almost exactly the same layout.

    We will insidously and slowly introduce changes as we discover the need for them while doing usability tests, specially cognitive walkthroughts. If it is needed by the user, we add it.

  2. Brian Lamb Says:

    Redesign for redesign’s sake is still happening witness GAP’s recent travails with their retail sites. Detailed story at Guardian unlimited,16376,1569784,00.html

  3. Loz Says:


  4. Jared Spool Says:

    Thanks for the grep, Loz. 🙂

    (As you can see, English is a second language for me. Baby talk was my first language.)

Add a Comment