Wednesday, 24 October 2007

Yes, you really should do Usability Testing!

You should do Usability Testing. Really. When you create a new website, Usability Testing should be part of your project.

If not, it will cost you more money and less people will use, visit and like you website.


Simply because you are not the user. It is really impossible for people in the project or even in the company or people in the web development business. You have no idea how the visitors of your website experience the site. An you surely will agree that you want them to like it and use it?

What's the occasion?

At the company I work now, we've done a redesign project of their new website. A Usability test was done (by another company) on the old site. I started to work on the project when it actually was started already. So in fact I was a bit too late. But I tried to make the project work by introducing User Centered Design (UCD) to them. And Usability Testing is part of UCD.

What I've done.

I couldn't do the whole UCD process, but I did some parts to catch up and ensure that the project would be a success. Since you may not know what the UCD process is, I will tell you what I did in this perticular example.

I made personas. Personas are fictional website visitors which fit the main profile of the visitors. This profile was determined by talking to the Business people about who their customers are and on which they target. I made 3 Personas.

The website was graphically designed by a design company who seemingly didn't have too much website design experience. They still designed for print instead of the web. And these are really different!

When we saw the first designs, I knew we had a lot of work to do... I won't get into detail here, but I made quite some remarks on Usability issues. There are a lot of things about the use of colour, contrast of colours, position of elements, what kind of buttons to use, the use of bread crumbs, and so on, you can tell in advance they won't work or confuse the user.

At the end, I've won on a lot of parts, and some parts were left as is. They probably would be exposed in an Usability Test!

Functional testing

Using the Personas, I made some scenarios with the Business people. The scenarios represented the most common and most important scenarios on the website. The developers of the site should consider if the Personas would be able to complete these scenarios. I sent the scenarios to the Business people and asked them to make it part of their test-plan. The company has a low maturity on testing, so it also stimulated them to make testing, in fact, functional testing, more important.

The testing process started a bit uneasily, but after all it was done quite well.

The Usability Test

After the site was developed and functionally tested, the company decided it should go live and then do a Usability test. Well, it would be better to do the Usability test first. But hey, as I told before, I got into the project a bit late, so it was a bit hard to change the planning for this.

And it wasn't too big a problem, since I could do the test right after the site went live.

No or low budget Usability Testing

I've done a cheap Usability Test. This meant that I invited some friends and acquaintances to participate in the Test. They've never done a test like this and they hadn't seen the website yet. I drove to them in my car, so that cut down on expenses.

When I designed the test, I used the Personas to pick the participants and the scenarios to perform the test. I used a demo version of a screen capture software package to record what the participants did.

I told the participants that they were not doing an exam, and that if something didn't work or was difficult, the website was to blame. I asked them to perform the scenarios and cleverly avoid any questions they had. I was an observer and nothing more. This simulated the situation they would be in when they visit the site from home or somewhere else without anyone around from the company itself!

I also had a questionaire about how they liked the site, what they would change etc. And finally, as a bonus, I gave them a sheet with keywords like "nice", "ugly", "easy to use", "family like", "professional", etc. and asked them which keywords fitted the website. The results of this can be used to match them with the ones the Business people would like to represent the company.

All went well and it took me about 24 hours all together to design and take the test with the 5 participants!

Last monday I made the presentation to present the test results to the Business people. And again, it really surprised me! Issues arose about which I haven't thought yet and offcourse also some issues which I predicted when the design company presented their design ;-).

Now what?

Doing a Usability test before you even develop the site, on a paper prototype or prototype, will reveal 85% of the Usability issues inyour design. You should do the test again when you have developed a first version. It will again reveal 85% of the Usability issues. After some iterations, when the site is live, you should continue to do Usability Tests. The science of Usability changes, and the way users behave does.

What will this cost?

Early Usability Testing is much cheaper. The costs of fixing the issues from an Usability Test in an early design stage are at least 5 times lower that when they are fixed after the site has gone into production.