Why Demographic Profiling Feels Wrong

Hannah Donovan from last.fm held a session called “Design as Service” at Zap Your PRAM last week. One of the things she talked about was designing for what she called “personas,” essentially made-up people with certain characteristics that a designer holds in their head when creating something.

I’d encountered a similar concept through a recent information architecture exercise we went through with a client and an IA consultancy: one of the things they had us do was to concoct three archetypal users that embodied some of the different characteristics that a possible user might — “Sally, a 42 year old mother of three children, works out of the home, technically savvy, drives a Volkswagen and travels twice a year” and the like.

Both in Hannah’s description of the process, and in the actual carrying out with our client it seemed rather odd and uncomfortable and imperfect, partly because people like Sally don’t actually exist (we’re all unique and special and cannot be typified, etc.), and partly because it seems vaguely like the sort of racial, ethnic and economic profiling that’s become insidious of late.

But mostly because it didn’t seem to remind me of the way that I actually create things.

And so it was with some pleasure that I read Tim Bray on this very topic (pointed to by John Bruber):

It’s like this: There’s only one person in the world whose needs and problems you really understand and whom you know exactly how to satisfy: that would be you. So build something that you use all the time, and, unless you’re really weird and different from everyone else, you’ve got a potential winner.

This resonates with me, and make me recall a non-digital example of this, an example that I used when asking Hannah a question on this issue.

Several years ago I was working for a local retailer. The previous manager of their shop was a woman who knew their customer intimately, because she was one of them: you had only to see inside her living room to know that the approach, the aesthetic and the product mix of the store reflected her tastes. And that was a good thing: the store had an attitude, a position, a editorial point of view (so to speak). It felt like more than the sum of its parts. Because it was: the manager had merchandised a store for people like her.

Later, when she’d moved on to other things, the store fell under the management of a much different group, of which I was one: middle-aged men who could only try to imagine what products, styles and merchandising would appeal to our customers. While we managed to get this right some of the time, we were, at best, making educated guesses and, at worst, relying purely on fictional stereotypes of what we thought our customers were like.

The store survived, but it has never been as good as it was in the very beginning because, as Tim Bray says:

Sometimes you can guess what people want, and you might get lucky. But probably not, so go ahead and build what you know for sure one person needs.


Chuck's picture
Chuck on October 24, 2008 - 21:16 Permalink

Personas aren’t supposed to be made up, though. They’re supposed to be amalgams of real people that you’ve actually met, and watched do their work. Done that way, they can be a very useful way of encapsulating the needs and, especially, the limitations of your end users.

In my work at SMART Technologies, personas were most useful as a reality check for overeager product owners. When you’ve done the research on the relatively limited computer skills of your target audience, and you’ve presented that research and convinced the development team and product manager, then embodying that finding in a persona helps create a touchstone for future reference: “Will Beverley be able to use this?” It also helps new team members come up to speed; you can point them at the relevant research if they want depth, but the persona makes an excellent shortcut.

Any consultancy asking you to create personas in a planning session is trying to get the benefit of the technique without laying the groundwork required. In that case I agree, it just won’t work (although there’s a minority heretical view that says the value in that case is getting everyone’s assumptions about “who the user is” on the table, even if they’re inaccurate).

My experience is that we can indeed be typified well enough for most interface designs. With UI work you’re not usually looking for small effects; you’re looking for huge, showstopping problems that are common to big swaths of people. Most folks over 40 develop some form of presbyopia; if you’re designing a website for that demographic, don’t set your type at 10px. Sounds obvious but you’d be amazed (actually, Peter, you probably wouldn’t be) how many sites violate such a simple rule.

Geez, I wish I hadn’t retracted my request to come to Zap! =)

Paul Lopes's picture
Paul Lopes on October 25, 2008 - 00:47 Permalink

Personas are useful for many reasons but I like to use them for one important reason: Personas make my clients think about their web site from their customer’s perspective. This is unique because most clients often think about their website from only their own perspective or that of their company, using terms or wording that is only familiar to them or assuming user actions based on what they would do — not what their customers would do. Personas help establish that a website is not built for the clients, but rather for the client’s customers and that the web site’s success depends entirely on this fact. Thus you don’t have to be a weirdo to benefit from personas.

Steven Garrity's picture
Steven Garrity on October 25, 2008 - 16:24 Permalink

Take personas and replace them with real people.

Stephen Pate's picture
Stephen Pate on October 25, 2008 - 19:09 Permalink

The demographic theory is hotly debated. My favourite authority on this says products serve people’s wifm’s not demographics.

What’s in it for me — wifm — is the filter used by all humans when presented by the 20,000 media messages a day. Hip hop song -hate hip hop, hop song so I move on.

WIFM is pretty basic — you want/need something, of a perceived value level for a certain price point. That’s what people buy products and services on.

What’s in it for men” Robin Woods, recommended by Gerry Goodis of “At Speedy You’re a Somebody” a slogan that made Speedy Muffler based on the untapped market of people who needed car repairs but felt like a doofus at the dealer/garage.

Interesting concept even if you don’t subscribe. Products for people — also used by Apple with all their products like iPod.