One of the main challenges in creating personas is deciding “where to draw the line”, i.e., making judgment calls of what information to include into the persona profiles and how many personas to create.
Ideally, we’d include all the information that the eventual persona users need and nothing else.
Ideally, we’d create as many personas as needed to represent the full range of user types in the dataset that we have collected.
But, as Chris Chapman says in his book (R for Marketing Analytics), just like there is an infinite number of slicing a pizza, one can create close-to-an-infinite number of different personas. Even if the two personas differ in the slightest detail, they are unique. An extreme example: two personas are the same except their hair color!
So, how big differences should we focus on? And how many personas do we actually need?
As the current body of persona research is unable to present patent solutions to these questions, they remain active and open challenges.
One direction that our research team has lately been investigating is shifting the making of these judgment calls from persona creators to the responsibility of their users. Namely, traditionally persona creators decide the number of personas and their content. However, there is no definitive reason as to why these decisions could not be made by the persona users themselves, as a part of the process using an interactive persona system.
An example of this philosophy is incorporated in one of our systems, Survey2Persona, that generates personas based on either the (a) demographic attributes or (b) survey statements that the persona user is interested in. So, essentially, with different information needs, different personas are generated. This corresponds to the idea of “slicing the pizza in different ways”.
Chris also discussed a related problem in his 2008 article, “Quantitative Evaluation of Personas as Information”. Namely, every attribute we add to the persona profile increases the informational complexity, i.e., we end up having more *possible* information configurations. If we want to cover all possible combinations, this means that the more information we want to convey with the personas, the more personas we need, and the number of needed personas can quickly reach thousands if we want to cover the whole available information space.
Therefore, there is a practical need of narrowing down the number, as persona users cannot, for cognitive limitations, manage thousands of personas at a time. However, the “inventory” of the thousands of personas could be ready in the database (or there is a readiness to produce any persona in realtime, on-demand) and the persona user can prompt the system to get a “slice” of the persona space based on what specific information aspect they are interested in. This is the idea behind combining the notion of personas with interactive systems.
Consequently, it means that no single persona generation or set is a definitive one, but there can always be subpersonas or alternatives to the personas a user is accessing at a given point in time. This thought is somewhat challenging, since often stakeholders just ask, “Show us our personas”, as if there would be one definitive list. This is because they have internalized the current paradigm as the only possible one; that there is data from which some specialists create a definitive list of personas that are “the” personas representing the customer base.
It is crucial, yet challenging, to learn away this static view of personas, replacing it with a dynamic view based on interactive use of systems based on current information needs. After all, what we need to know about our customers today is not the same than what we need to know tomorrow; therefore, it makes sense to understand that personas are “perishable” in nature; we don’t need to print posters of them in our office walls, but we can “call” them for a given decision making situation and then move on.
Coincidentally, adopting this new philosophy of seeing personas as perishable and expendable can yield positive effects for inclusive design. Namely, while it is unlikely that fringe or marginalized user groups end up being included in a “traditional” persona set of 3-7 personas because the “seats” in this persona set are typically reserved for the average or most typical users, the flexibility of slicing the user data in unique ways increases the possibility of marginalized user groups emerging in the persona set, which in turn increases designer awareness of these groups’ attributes, thereby promoting inclusive design.
Therefore, letting go of the notion of “one static persona set” can also help fight against harmful stereotyping.
Any thoughts? Shoot me an email if you have!