# Why Your Personas Are Wrong

Wrong for UX. Wrong for America.

Meet Norma. She’s a 23 year old, middle class American female. She makes \$50,000 per year and has 2.5 kids. She uses Facebook and Pinterest. Her hobbies include reading, gardening, and spending time with her family.

Except that none of those things are true. Like most UX personas, Norma is a bunch of fictitious details draped over an empty frame of averages.

In the 1940’s, Norma was kind of a big deal. She was a sculpture created by gynecologist Robert L. Dickinson based on measurements taken from thousands of patient examinations. The local press ran a contest to find the Ohio girl who most closely matched Norma’s dimensions.

They thought it’d be an easy quest. After all, Norma’s whole thing was her averageness. She should’ve been 900,000 in a million, right?

Of the 3,864 women who submitted their measurements to the contest, exactly ZERO of them were perfectly average in every way.

### The Demagoguery of Demographics

People often mistake “average” for “typical.” As it turns out, it’s not that unusual to be an outlier in something. Most people are atypical on at least a few attributes. In trying to describe an “average” user, we end up creating personas that don’t represent anybody.

Instead, we should provide our teams with real data that capture the full range and diversity of our users. To see how, you’ll have to understand a few things about the bell curve…

It’s true that very few people are exactly average, but you can expect 68% of people to be within one standard deviation of the average. That’s the dark blue region in the graph above. It’s an essential stat.

When the standard deviation is high, the average is worse than useless. It’s misleading.

The range is simply the difference between the largest and smallest values. You can represent both the range and the standard deviation with box plots like these…

You may want to put some actual, numerical values in there, of course, but even the shape of a box plot conveys information. It tells you how diverse your users are and whether they tend to skew high or low.

Instead of reporting “our users have an average income of \$50,000,” you should say something like “our users’ incomes typically range from \$15,000 to \$85,000 with a long tail of highly affluent users.”

Rather than give your persona a bachelor’s degree, you should report that “users’ educational attainment typically ranges from high school graduates to master’s degrees.” You get the idea.

This isn’t just a matter of capturing diversity. For any dynamic attribute (intelligence, coordination, computer literacy… all the useful things), even a single person’s behavior can vary dramatically from one moment to the next. Even a genius will act like an average person a lot of the time. Standard deviations between people can also tell you a lot about the range of variation within individuals.

All this detail helps you, and your teams, avoid stereotypes when thinking about your users. Stereotypes are the enemy of empathy. Demographics can only take you so far, however. Real empathy only comes from understanding your users’ needs.

The astute observer may have noticed that the methods above don’t lend themselves to traditional personas, ones with cute nicknames, stock photos, and made-up hobbies. That’s for the best. At their worst, those fictitious details are just more stereotypes.

Instead, I recommend building your personas around need-states. It’s a marketing term, but I’m sure you can handle it. Need-states are the reasons people use your website or app. They’re the problems that you can help people solve. In short, they’re why your product exists.

Aside from making money, of course.

A single person, with a single demographic profile, can come to your app in many different need-states. In fact, it’s common for people to progress through a series of need-states as they interact with an app. For example, people interact with content websites in a sequence of two need-states: Evaluating and Consuming.

Users typically begin their experience with a search. They click through to a piece of content and try to make a quick judgement about whether or not they’ll like it. They look at the author, source, social proof, and other indicators of quality.

In this context, social proof is anything that signals the endorsement of others: ratings, reviews, view counts, like counts, and so on.

In this need-state, users have very little patience. They’re not going to be receptive to ads or calls-to-action (e.g. “Sign up for our newsletter!”). If they decide the content isn’t what they’re looking for, they’ll bounce right back to Google or wherever else they came from. Optimizing for this need-state means hiding most of the content and pretty much anything that counts as monetization.

Obviously, that can be a hard pill to swallow. It took many years for online publishers to get savvy about this. It’s only after you’ve convinced an Evaluator to read something that you can convert them into a Consumer. You’ll want to adjust your page layout to display the full content, fire up the ad server, and get those calls-to-action in front of some eyeballs.

The challenge for designers is how to lay out a page in a way that transitions smoothly from one need-state to the next. To get there, of course, you first had to identify those need-states and no mountain of demographic data or list of made-up hobbies was going to get you there. It’s not enough to know who your users are. You also have to know what they want.

Here’s how I might write all that up for a hypothetical client…

Demographics

Our content is intended for people who are the primary caregiver for an elderly parent. Though their average income skews a little low, they come from a wide range of socio-economic backgrounds. We should avoid making assumptions about “coupon-clippers” and “cash-strapped retirees.”

Unsurprisingly, our online content tends to reach people who use the internet a lot. We should focus our marketing efforts on reaching potential users in other media. We should focus our design efforts on making the site more familiar and less intimidating for low-frequency visitors.

Need-States

1. Evaluator: “I want something interesting to read.”
• Trying to make a quick judgement about the quality of the content and whether or not they’ll like it.
• Likely to bounce if they decide not to read the article, but might click through to related content.
• Not receptive to ads or calls-to-action.