“Everything should be made as simple as possible, but not simpler” — attributed to Albert Einstein
Designers talk a lot about simplicity and parsimony in what we produce. This is a common refrain across disciplines — visual design, interaction design, service design, and many other domains — but we talk less about this principle in the tools we use for solving design problems. Many designers use personas to bring their understanding of users into their design work. A persona is an amalgam of people seen in design research, an amalgam of a type of potential user seen in the field, fictionalized and abstracted enough that the designers don’t get too hung up on the irrelevant idiosyncrasies of particular individuals.
I worked in the Cooper studio with Alan Cooper, who coined the term “persona” in the early days when we were developing them in our practice. One part of that practice I don’t think we talk about enough is how to build out a set of personas to describe the shape of the user population when we have found varying types of people. As Einstein would have it, we aim to limit ourselves to as few personas as possible — but no fewer than we need. So how do we know how many that is?
Think of personas like tentpoles under a circus big top tent. Users stand under the tent. Designers want to cover as many people as possible with as few tentpoles as possible, giving people plenty of headroom; the further from the pole people stand, the less design which targets that tentpole — that persona — will serve them. Generally a designer building personas places those tentpoles somewhere in the middle of clusters of people; no one person is just like the persona, but enough people are similar enough that targeting the design at the persona addresses all of those people. Often observing users makes this easy; one can see a few clear clusters of people.
But in other situations, the structure of the user population requires some art to address. One may find an array of users such that it doesn’t make sense to place your persona in the middle — that leaves too many people poorly represented — but if one plants a tentpole near one end of that cluster and another tentpole at the other end, it covers the whole range. A pair of personas may thus represent not distinct types in the population but rather tendencies among an unstructured space of users.
The “geometry” of the user space can be trickier still, such that choices about how to frame the user population as personas require artfulness and hard choices, especially in pursuit of keeping the persona set “as simple as possible but no simpler.” This metaphor is getting a little abstract, so consider a specific example. I once worked on a project which included people from various roles in the organization communicating through a system. We knew we had to address IT, and we started the project assuming that we would have a persona representing IT professionals. But after field interviews and synthesizing our research, we found they differed a lot.
For example, they had different attitudes toward vendors. Some were boosters for particular vendors, building relationships with their reps and accumulating training certifications in their tools. Others were skeptical about vendors, fighting against developing too great a dependency on any one of them, even preferring to roll their own solutions despite it creating greater cost. Some were specialists, some were generalists. Some were cool pragmatists who accepted the ups and downs of their work with aplomb, some were dryly cheerful cynics rich in funny stories where they got to say I Told You So. These characteristics existed in all different permutations. It was too much variation to represent with one persona.
In our user research, we talked to IT professionals who fit into a bit of a stereotype of Middle Aged Guys With Beards; those guys tended to be skeptical of vendors, more generalist than specialist, and dry cynics. Even though no one person we spoke to was all those things at once, it was enough to hang a persona on. So we had our personas Will and Viola. Will had the beard and loved Linux, and Viola liked Windows NT (because this was some time ago, and nobody ever loved NT). Two personas for one role. With those two, we could address a range of people who differed across several axes.
This example also hints at how the reverse can happen: one persona to address a few different roles. Maybe a real estate broker is a good persona to represent a professional who handles a lot of paperwork on the go, and so can also stand in for a salesperson, an insurance inspector, et cetera. This looseness in the relationship between how personas are defined and the roles people fill reflects how designers should not define personas in terms of the tasks people do but look a layer deeper to the goals they pursue: what they try to accomplish, and how they think about it.
So how many personas belong in the persona set which represents the users in a domain? There is no magic number; it depends on that domain. In a consumer space with a broad range of people, it may turn out that people have a range of widely divergent goals, requiring several personas, but in many cases even a big population of people have such similar fundamental needs that only one or two well-conceived personas will serve. Alan Cooper liked to talk about how the inventor of the rollaboard suitcase was a pilot who was thinking about the needs of people on flight crews … but that single design persona produced a solution that suits most people who fly.
How many personas do you need to design your product? As few as possible … but no fewer.