I'm listening to Malcolm Gladwell's talk at Pop!Tech. I'm blogging in real time as I listen, so this is straight from my head to the blog.

A chair maker decided to use mesh that allowed the chair to breath and yet does not pinch, as well as a bunch of other cool innovations. (I had one of these at my previous job--they're great!)

Anyway, the chair has to be comfortable and look good. It was very comfortable--an 8 out of 10. Still people didn't like the way it looked--in fact they thought it was ugly. Market testing, for this aspect, was abysmal! Even designers weren't keen on it. People said it looked like lawn furniture!

The catch is that it's the best selling chair in the history of chairs--the Aeron. Eventually, people decided it wasn't ugly. And then, it started winning awards. And then, everyone loved it! Go figure.

So, it went back into focus groups. And now the looks were up in the 8 to 9 rating range for looks as well as comfort. Huh?

Well, we try to come up with things that people like. So, we ask them. And they tell us. But there's a problem--something is lost between the feeling of a preference and an expression of a preference.

If Herman-Miller had listened to this research, they'd have passed up the greatest success in their history. So, what gives?

Thing is, preferences are unstable. They are not written in stone.

Take, as another example, New Coke. The Pepsi Challenge made Coca-Cola rethink their whole product--they made it more like Pepsi. And they did hundreds of thousands of tests on this--and New Coke came out on top.

Ha! Preferences weren't stable. And the root of New Coke's failure is that the old blind taste test sucks. There's something wrong with a sip. No one takes a sip when drinking! One sip makes people prefer the sweeter, Pepsi-like drink. But actually drinking in a "home use" test leads people to very different conclusions.

The triangle test illustrates this. Give three glasses of cola, maybe with two Pepsis and one Coke. People will be able to pick out the Coke only 33% of the time, same as chance! This despite the fact that the "Pepsi Challenge" results seem accurate.

When people tell you what they think, they're telling you stories. They pick up unconscious cues in testing scenarios, often outside of the test, that we then make stories about that may have nothing to do with the cues we've seen. In fact, when people explain what they're feeling, the stories may bear no relationship with reality.

Here's the Perils of Introspection problem: making people explain what the want, changes their preference and change it in a negative way. Why do people, when ask to explain a preference, gravitate to the easiest choice to explain? Because people's language doesn't allow them to explain the more sophisticated, more daring choice, and go with the easiest to explain.

So... can we trust people? Maybe. There are a class of products that are tough to interpret. And people will just call it ugly and say they don't like, when, in fact, they just can't explain it. So we when use what people say to determine what we ought to do, we may throw the things that are most interesting and useful!

People who come with new ideas need to be careful in how they interpret the evidence from consumers. We most be skeptical of negative responses to things that are for people to explain their feelings about.

Also, although we've made great strides in understanding cognition, we don't entirely understand our own hearts.

Thus spake Malcolm.