[Sigia-l] User satisfaction vs. dissatisfaction ratios for unsolicited feedback

[George Olsen]

>>I believe it has been shown that unhappy (unsatisfied) users will
>> provide negative feedback far more often than happy (satisfied) users
>> provide positive feedback for a web site.

Unsolicited feedback is like letters to the editor, only those with strong
views are typically motivated to offer feedback -- and usually it's
negative feedback.

That said, in other situations it's often a good idea to measure
satisfaction and dissatisfaction separately. (A technique known as Kano
analysis in survey research.) Essentially you ask users how satisfied they
would be if they can accomplish a particular task (drawn from your
scenarios), or if a certain feature is present, etc. Then you ask how
dissatisfied they'd be if it were absent, didn't work, etc.

The idea is there are some things that are assumed to work right -- like
the tires on your car -- and while people won't notice them if it does
work right, they'll be upset if it doesn't.

In other case, an absence of a feature won't be a negative, but if it's
present then it generates a lot of "unexpected bonus" satisfaction. In
fact these, when you get these, they're often the great selling points.

Then there are features whose satisfaction/dissatisfaction perform in a
pretty straight-line way, i.e. the better/worse they are, the more people
are satisfied/dissatisfied.

Finally, there are things people are just indifferent about -- people may
_say_ they want it, but it really doesn't influence how they feel about
the overall product.

Needless to say, this can be an extremely useful way to prioritize
features. I haven't had a chance to do it myself, but people I know in
product development first use user research plus personas and scenarios to
develop a list of potential user tasks and goals the product could enable,
and then use a Kano survey to reality check and prioritize the concepts
with a wider sample of the target users.

While running a survey obviously adds cost, I think it's a useful
complementary technique, since the risk in qualitative research is that
the small number of people typically involve could accidently skew your
results (i.e. you happened to get some people who were atypical).

_____________________________________________________________________
George Olsen                           george at interactionbydesign.com
User Experience Architect                                310-993-0467