Entries Tagged 'Website Usability' ↓

Until recently it cost thousands of dollars a month to subscribe to sophisticated testing services that would let you gather empirical data about proposed website design changes.

Now there’s Google’s Website Optimizer. This free, powerful, simple tool lets even the most analysis-phobic website owners test and evaluate proposed changes to their websites.

I attended a presentation at Google tonight (gotta love living in Fremont) by Tom Leung, the product manager for Website Optimizer. As always, I listened with an ear for tidbits useful to independent professionals and solopreneurs.

What Website Optimizer Does

Website Optimizer lets you test changes to your website. For example, say you want to change your newsletter sign-up box and you and your webmaster have different ideas about how it should look. Instead of discussing the change ad infinitum, just test it. With Website Optimizer, you add a little Javascript to your page and Google then shows half of your visitors one version and half the other version. Website Optimizer then measures which version results in more newsletter sign-ups. Simple as that.

Google uses this tool all the time to fine-tune their websites. If it’s good enough for Google, it’s good enough for you.

Google Website Optimizer for Independent Professionals

Website Optimizer lets you do simple A/B testing (like the newsletter sign-up example above), or you can do sophisticated multivariate testing on a bunch of webpage elements at once. I don’t recommend the latter for independent professionals since accurate multivariate testing requires more website traffic than our websites are likely to get.

Tom offered a couple of rules of thumb for testing. You need at least 100 “conversions” (website visitors taking the action you want) in order to have your test be statistically valid. He also said that for small sites, you want to make sure you have at least 500 page views a month before you even think about doing any testing.

Some Lessons Learned

Tom and his crew have been working with this tool for a few years now and identified a few take-home points:

• Website testing is a continuous process. You’re never really finished. There’s always something else you could/should be testing.
• You need compelling content, enough website traffic, and the discipline to make tough decisions about what is worth testing and what isn’t.
• Keep your testing simple. This is particularly important for small, low-budget sites.
• Finally, and most importantly, realize that this is an arms race. Whoever has the best-converting website will be able to pay more for leads acquisition from AdWords and similar programs. This excellent article, Google Website Optimizer 101, shows how to leverage your gains from your improved conversion rate.

On a personal note, I’ve got to say that I love this tool. I’ve been trying for years to genuinely divorce myself from attachment to my crackpot idea du jour, and this is just the ticket for that.

Steve Krug, a usability expert, gave a presentation in Seattle tonight on web site usability entitled, “Yes, Virginia, There Is A Perfect Web Page.” He says usability comes down to two main things:

1. Use effective “You Are Here” indicators to let readers know where they are in a site. These should be ubiquitous and consistent, and they should be LOUDER than parallel but less important elements on the page. See the way StumbleUpon uses “live” tabs that are brighter than other navigation choices and that blend into the main content background color for a good example of what he’s talking about.

2. Use prominent, well-placed page titles to let users know what the page is about. These titles should, of course, be consistent with the link you clicked on to get there (e.g., if you click on a link that says “Green Blocks” you should land on a page with a prominent title that says “Green Blocks”).

That’s it. Use those two elements consistently across your web site and your site visitors will happily navigate your site.

In the “Stump the Chump” part of the session, he took a look at my new web site, Bodywork U, which he found no major flaws with (though I did leave the presentation with a to-do list of ways to better apply his advice to the site).

In the Q&A session after his presentation, someone asked about the best way to use breadcrumb trails. He said that any one page on a site should have only one breadcrumb trail and that if you link to it from some other place on the site think of it as a cross reference. I love this observation. For one thing it’s the way I’ve always used breadcrumb trails. For another, I love that this now well-established usage works differently than the source of the analogy - that is, a real breadcrumb trail would be a list of the pages you had visited previously to get to the current page.

I also made a small-world connection with Steve after the session. He had mentioned that he went to Boston College, and I know a bunch of BC alumni from my East Coast days, so I asked and, what do you know, it turns out he’s good friends with my pal Jim in Boston.