I love this article by Amy Hoy on why logic fails to predict what people will buy. Here’s why she’s right, and a way you can model and predict behavior.

Here in the land of high-tech, I see it again and again: Logical, sensible people using logical, sensible arguments to predict what will sell — despite repeated evidence that logic and sense are not why people buy.

I have to confess. This was me:

“You probably didn’t believe anyone would pay $399 for an MP3 player that couldn’t even hold half as much as the Creative Jukebox. You probably knew the iPhone would flop because it didn’t have third party apps, 3G, GPS, multi-tasking or even friggin’ copy and paste.

“You probably thought the iPad was ugh, just a big iPhone, who cares.”

Almost. I was wrong about the iPod and the iPhone, for exactly the reasons she cites here. But when the iPad came along, I had learned my lesson and redeemed myself, winning a bet with an engineer friend who knew, absolutely knew, it would fail.

There is something I do know about who will buy the Apple Watch: it won’t be me, at least not for this year’s product (just as I did not buy the 1.0 versions of the iPod, iPhone, or iPad). That’s the point: I can predict success of a product, but only if my customer happens to be me.

David Packard called it “the next bench syndrome.” In HP’s early days, the company’s engineers could design by asking the guy at the next lab bench what he wanted, because they were designing products that electrical engineers would be using. When they began developing business computers, that model began to fail.

So how do we know what customers will buy? Well, there are some expensive methodologies that work well for established markets but most of my clients have something more like Apple’s market — new markets for new services and products that won’t predict well. And they don’t have Unilever-sized budgets for research. High-tech firms tend to rely on intuition and some sense that they know the customers. Risky business.

You can only know your customers by their actions

Amy Hoy says, “You can only know your customers by their actions.” But how do you know them all if they’re not you? A methodology that works for many of my clients is persona development.

I define a persona as a customer type based solely on desires and behaviors. We use the company’s inside knowledge to discover a very small set of personas (typically 3-5). The goal is to define them so clearly and concisely that everyone in the company can know them by heart. This is not the same as customer segmentation, which can be complex and detailed. Segments are for automation and procedures; personas are for people. We refine the personas through a series of processes and do small research projects to verify anything that’s uncertain. We create a chart showing each persona’s problems, emotional drivers, what products we have for them, benefits and unique value propositions, competitors, etc. The chart is detailed enough to capture the customer base’s desires but simple enough that everyone can know them.

With carefully designed and researched personas, a company has a fighting chance to know what they will want and make decisions based not on what people in the company would want, but what their real customers and prospects would want.