Personas have come under fire for including extraneous information that, beyond simply being unnecessary, can distract or contribute to misunderstandings about users and their contexts. For instance, Alan Klement suggests that by describing a set of attributes, personas distract teams from the actual problems their users are hoping to solve.
Much of the critique of personas has centered around their inclusion of demographic information and photography to represent the user. Indi Young and Klement have, in their own ways, suggested that including demographic information can instigate subconscious reactions and biases in team members. They may therefore inspire "anti-empathy." At best, demographic data may provoke divergent understanding as different members of the team interpret even soft descriptors like "low-income," "middle-aged," or "college student" in different ways. This is not to say that there are never situations when this kind of information is useful; but, these are likely exceptions rather than the norm.
Young, Klement, and others have therefore suggested adjustments to personas to mitigate the risk of biases. Much of this advice suggests removing demographic information or any data that would align the persona to a broad population rather than an archetype of an individual. For instance, Young suggests using gender-neutral names. However, because even those names may be suggestive of culture, race, or class, it may be preferable to refer to the "personas" with general descriptors rather than names at all. Young has also proposed using demographic data to disrupt assumptions people might have of a given group.
Personas have also been criticized for being too reductive. This is, to some extent, unavoidable; personas are by intent and by design meant to condense research into an accessible format. Spool says that personas can be a valuable tool, but suggests that most teams create overly simplistic user models that don't actually add to the process. "They look good," he writes, "but they get ignored."
It would be impossible to fully capture the complexity of every user and their context in a diagram that can easily be digested by a number of individuals. However, this creates the risk that the design team will fill in gaps with their own assumptions about the user. Klement writes that "[b]ecause personas focus on creating a story made up of a customer's attributes, instead of a story that explains a purchase, personas leave the brain in an unsatisfied state." This leads each member of the team to come up with their own, potentially different, story about the persona's needs and motivations. Teams should take care, then, to consider carefully whose experiences are represented by the personas and whose are left out.
Personas may also become too strong a stand-in for actual user research. That is, once personas are established, they may be used as an excuse for not engaging in further contact with users. This is dangerous. Even the most expertly designed persona can provide only a rough approximation of the real lived complexity of another human being. We can't test designs with persona, nor can we gather meaningful insight from them. We need real people for that. A persona shouldn't be the end of research. It's represents just part of it the research work to be done.
Better than any persona is ongoing exposure to actual users.
Klement, Alan. “Replacing Personas With Characters.” Medium (blog), September 2, 2017. https://medium.com/down-the-rabbit-hole/replacing-personas-with-characters-aa72d3cf6c69.