As previously discussed, personas are derived from patterns observed during interviews with and observations of users and potential users (and sometimes customers) of a product. Gaps in this data are filled by supplemental research and data provided by SMEs, stakeholders, and available literature. In constructing a set of personas, we are looking to segment use across a set of
Creating believable and useful personas requires an equal measure of detailed analysis and creative synthesis. A standardized process aids both of these activities significantly. The process described in this section, developed by Robert Reimann, Kim Goodwin, and Lane Halley, is the result of an evolution in practice over the span of dozens of interaction design projects.
Revisit the persona hypothesis.
Map interview subjects to behavioral variables.
Identify significant behavior patterns.
Synthesize characteristics and relevant goals.
Check for completeness.
We will discuss each of these steps in detail in the following sections.
After you have completed your research and performed a cursory organization of the data, you
List the complete set of behavioral variables observed. Demographic variables such as age or technical skill may also seem to affect behavior, but be wary of focusing on
If your data is at variance with your assumptions, you need to add, subtract, or modify the roles and behaviors you anticipated. If the variance is significant enough, you may consider additional interviews to cover any gaps in the new behavioral ranges that you've
After you are satisfied that you have identified the entire set of behavioral variables exhibited by your interview subjects, the next step is to map each interviewee against each variable range that applies. The precision of this mapping isn't as critical as identifying the placement of interviewees in relationship to each other. In other words, it doesn't matter if an interviewee
Figure 5-4: Mapping interview subjects to behavioral variables. This example is from e-commerce. Interview subjects are mapped across each behavioral axis. Precision of the absolute position of an individual subject on an axis is less important than its relative position to other subjects. Clusters of subjects across multiple axes
After you have mapped your interview subjects, you see clusters of particular subjects that occur across multiple ranges or variables. A set of subjects who cluster in six to eight different variables will likely represent a significant behavior pattern that will form the basis of a persona (Goodwin, 2002). Some specialized roles may exhibit only one significant pattern, but typically you will find two or even three such patterns.
For a pattern to be valid, there must be a logical or causative connection between the clustered behaviors (Goodwin, 2002), not just a
For each significant behavior pattern you identify, you must synthesize details from your data. Describe the potential use environment, typical
Brief bullet points describing characteristics of the behavior are sufficient. Stick to observed behaviors as much as possible; a description or two that sharpen the personalities of your personas can help bring them to life. Too much fictional, idiosyncratic biography, however, is a distraction and makes your personas less credible (Goodwin, 2002a). Remember that you are creating a design tool, not a character sketch for a
One fictional detail at this stage
important: the personas' first and last names. The
It sometimes makes sense for the set of personas for a product to be part of the same family or corporation and to have interpersonal or social relationships with each other. The typical case, however, is for individual personas to be completely unrelated to each other and often from completely different geographic locations and social groups.
It makes sense for personas to have social relationships between each other only if:
You did not observe any behavioral variations in your interview subjects
Doing so is critical to
If you create personas that work for the same company or have social relationships with each other, you might run into difficulties if you need to express a significant goal that doesn't belong with the pre-established relationship. While a single social relationship between your set of personas is easier to define than several different, unrelated social relationships between individual personas and minor players outside the persona set, it's much better to put the initial effort into development of diverse personas than it is to risk the
Goals are the most critical detail to synthesize from your interviews and observations of behaviors. Goals are best derived from an analysis of the group of behaviors comprising each persona. By identifying the logical connections between each persona's behaviors, you can begin to infer the goals that lead to those behaviors. You can
To be effective as design tools, goals must always directly relate, in some way, to the product being designed. Typically, the majority of useful goals for a persona are
. You can expect most personas to have three to five end goals associated with them. Life goals are most useful for personas of consumer-oriented products, but they can also make sense for enterprise personas in transient job roles. Zero or one life goal is appropriate for most personas. General experience goals such as "don't feel stupid" and "don't waste time" can be taken as implicit for almost any persona. Occasionally, a specific domain may
At this point, your personas should be starting to come to life. You should check your mappings and personas' characteristics and goals to see if there are any important gaps that need filling. This again may point to the need to perform additional research directed at finding particular behaviors missing from your behavioral axes. You might also want to check your notes to see if there are any political personas that you need to add to
If you find that two personas seem to vary only by demographics, you may choose to eliminate one of the redundant personas or tweak the characteristics of your personas to make them more distinct. Each persona must vary from all others in at least one significant behavior. If you've done a good job of mapping, this shouldn't be an issue.
By making sure that your persona set is both distinct and complete, you will be able to maintain a manageable set of personas.
Your list of bullet point characteristics and goals point to the essence of complex behaviors, but
A typical persona narrative should be no longer than one or two pages of
The narrative must, by nature, contain some fictional events and
Be careful about precision of detail in your descriptions. The detail should not exceed the depth of your research. In scientific disciplines, if you record a measurement of 35.421 m, this implies that your measurements are accurate to .001 m. Likewise a detailed persona description implies a similar level of observation in your research (Goodwin, 2002a).
When you start developing your narrative, choose photographs of your personas. Photographs make them feel more real as you create the narrative and engage others on the team when you are finished. You should take great care in choosing a photograph. The best photos capture demographic information, hint at the environment (a persona for a nurse should be
By now your personas should feel very much like a set of real people that you feel you know. The final step in persona construction finishes the process of turning your qualitative research into a powerful set of design tools.
All design requires a design target—the audience upon whom the design is focused. The personas, which we have created, represent the possible candidates for that design target. A single interface can only be designed for a single persona.
Design each interface for a single, primary persona.
What we then must do is
our personas to determine which should be the primary design target. The goal is to find a single persona from the set whose needs and goals can be completely and happily satisfied by a single interface without disenfranchising any of the other personas. We accomplish this through a process of designating persona types. There are six types of persona, and they are typically designated in
We discuss each of these persona types and their significance from a design perspective in the following sections.
Primary personas represent the primary target for the design of an interface. There can be only one primary persona per interface for a product, but it is possible for some products (
A primary persona is not satisfied by a design targeted at any other persona in the set. However, if the primary persona is the target, all other personas are at least
Choosing the primary persona is a process of
Sometimes a situation arises in which a persona would be entirely satisfied by a primary persona's interface if one or two specific additional needs (not required by the primary) were addressed by the interface. This indicates that the persona in question is a secondary persona for that interface, and the design of that interface must address those needs without getting in the way of the primary persona. Typically, an interface will have zero to two secondary personas. More than that again indicates scope problems with the product.
Customer personas address the needs of customers, not end users, as discussed earlier in this chapter. Typically, customer personas are treated like secondary personas. However, in some enterprise environments, some customer personas may be primary personas for their own administrative interface.
are somewhat different from the persona types already discussed. They are not users of the product at all; however, they are
directly affected by the use of the product
. A patient being treated by a radiation
Like served personas,
aren't users of the product. Unlike served personas, their use is purely