[Sigia-l] length of nav labels

[Jared M. Spool]

At 10:15 PM 8/9/2005, Listera wrote:
>Boniface Lau:
>
> > When the numbers come with supporting hypothesis, one can check
> > whether the hypothesis applies to the current design context.
>
>I'd go even further: compared to what?
>
>It's usually said, "This tests well." Compared to what? What other mildly or
>radically different approaches was it tested against? What was the context?
>And most importantly, what was the design goal?

Well, I can't speak for others, but when I say, "This tests well," I'm 
talking about how it compares relatively to other designs, either 
alternative prototypes of the same site/application or competitive designs.

Our tests draw from a collection of measures to determine their relative 
score, such as task completion, brand engagement, satisfaction, revenue 
generated, time-to-completion, errors, deviations from ideal path, 
subsequent activity, and impulse purchases. We choose specific measures 
from requirements set out by our clients to satisfy their business objectives.

So, a design "tests well" when it is more likely to succeed at business 
objectives than other designs we tested.

 From this, we've learned that certain design components consistently show 
up in the designs that test well. Other design components consistently show 
up in designs that don't test well. And others just aren't consistent. 
We've learned that designers are happy to learn which components are in 
which group.

>Terrence Wood:
>
> > Surely, whitespace is the most valuable asset in a designers toolbox,
> > and the lack of the same is a tell-tale sign as to whether an actual
> > designer has had involvement in the implementation of a web site.
>
>Yes, but folks in the usability business don't care much for or truly
>understand design. To vulgarize the matter, for example, take the triumph of
>usability culture:
>
><http://www.mcmaster-carr.com>
>
>OK, you might be saying, please take it! :-)

I told you that designers hate it.

I'm not sure I'd suggest that this site is a triumph of usability culture. 
(I'm not even sure I understand what that phrase means.) Most people who 
aren't in the construction materials industry look at this site and wonder 
how anyone can use it.

I think it's only viewed as a triumph within its specific industry segment 
(and my office).

>I have no idea what the design goal here was. Jared says it "tests well."
>Compared to what? The speed with which a user can find something,
>statistically? That's like claiming organized crime families are a model of
>nurturing for kids because they foster loyalty among the peer group. OK, but
>compared to what? At what cost? What other opportunity costs are we talking
>about here?

For an e-commerce site (which this is), the design goal is to sell product 
(both short- and long-term).

In the case of McMaster Carr, this design has dramatically increased 
revenues to the site. It has increased brand engagement. Customers prefer 
it to other competing designs (such as http://www.grainger.com ) and decide 
to do business accordingly. The folks at McMaster have been refining it for 
years and this refinement succeeds best for their organization.

So, when I say it "tests well", I mean that it performs relatively well 
compared to other designs we tested when measured for these goals.

>I'm sure at some point the Yahoo Directory tested well too, in the absence
>of Google as another approach!

Absolutely. I'm in complete agreement. Which is why I'd never say that 
McMaster Carr is a best practice. ("Best practice" is not a term we ever 
use in our work.)

I would love to see a design that tests even better than this one. And when 
I do, I'll be showing them to anyone who'll listen, just like I keep waving 
the McMaster-Carr site around.

And you're right: I don't understand design. I look to see what performs 
well in our tests. Things that are "designed well" often don't perform well 
and things that are "vulgar abominations" sometimes test exceptionally 
well. For many, good design doesn't match up with good performance.

I like the McMaster-Carr site *because* it so clearly goes against what 
people believe is good design. It makes us all question what we're trying 
to accomplish when we create something. I rarely suggest a client mimic 
their design (we've seen it works very well for them with their users, but 
rarely translates well to other contexts), but I like how it makes us think 
twice about what's good and what's bad.

My clients aren't so interested in good design as in good performance. I 
acknowledge that there is probably more to design than meeting performance 
measures -- I just don't know what those elements of design are.

Asking these questions and having these discussions are good.

And in the end, you're gonna design what you want to design. If, in the 
process, I've made you step back and think for a moment, then I've 
accomplished my goal.

Jared

Jared M. Spool
User Interface Engineering
http://www.uie.com    jspool at uie.com

UI10 Spotlight Presenter: Flow author Mihaly Csikszentmihalyi
October 10-13, Cambridge MA. See details at http://www.uiconf.com