• Avoiding Half-baked Personas

    I’ve written about the tension between truth and fiction in personas before. In that tension is the power of personas as a design tool but it is also their greatest potential weakness. Too much fiction leads to misguided design. Too little fiction leads to uninspired design. The magic stuff that gives us that proper amount and kind of fiction is intuition. It is the leaven in the bread; the spark of real live human that we put into the data analysis.

    “It is through science that we prove, but through intuition that we discover.”

    -Henri Poincare

    “There is no logical way to the discovery of these elemental laws. There is only the way of intuition, which is helped by a feeling for the order lying behind the appearance. ”

    -Albert Einstein

    The key to good persona work is understanding the strengths and dangers of intuition at different stages of the research and design process. I mention stages and process because we have to acknowledge that personas are never strictly done. Good ones are always evolving in response to changes in the organization using them, the market they compete in, or even the particular project they are used to tackle. This raises important questions: when is a persona done enough to be useful? what’s the right ratio of data to intuition at any given stage? Many people have written about personas in their more final stages but there’s been less talk about the interim steps. All this came up recently during a discussion on our internal Adaptive Path mailing list about “ad-hoc personas” and “proto-personas.”

    Note: some of you may not have heard of these terms or, like some people in Adaptive Path and in the field, might have heard them used interchangeably. I’d like to use this post to try to draw some distinction for the sake of clarity in the conversation.

    Ad-hoc personas: the danger of half-baked intuition

    “Ad-hoc personas” is a term generally used to refer to personas based on what an organization happens to know right now without a new data collection effort. (“ad-hoc” = “improvised or impromptu”.) Typically, they are based mostly on intuition and unconnected bits of market research. (Hence, Jared Spool doesn’t even want to call them “personas” but just “user descriptions.”) These kinds of descriptions can be helpful, especially in a better-than-nothing situation or as the first step in a longer process. Or as Chris Risdon said…

    “Ad-hoc personas are great when developed as a starting point by an ‘in-house’ UX team that can invest in personas that evolve and mature over time as you do iterative small-n, evaluative research (such as in an agile development environment). They serve as a starting point so everyone can start with a shared understanding of the user, even if flawed or incomplete (otherwise everyone has their one flavor of who the user is). These personas mature with the addition of new data with each iteration.”

    But that assumes that you will be revisiting and iterating. Not all organizations can or know that they should do that. It’s important not to focus too much on these half-baked personas because they have serious potential problems. As Kate Rutter points out…

    [Paraphrased] Personas should *always* be based on research findings. Research unfogs the internal goggles and is a breakthrough way of seeing people for who they really are and meeting their real needs. Ad-hoc personas can feel a little like putting the fox to mind the chicken coop. What people think their constituents are like can be quite far from what actual constituents think and how they behave. Therein lies much madness. Although these are helpful tools when there’s simply no alternative, I’m concerned that they can become a crutch to real and nuanced understanding of the audience needs for products. Honestly, ad-hoc personas are more of a tourniquet than a deliverable…an emergency-based procedure designed to stop the bleeding until real care (and real research) can take place.

    Proto-personas: the power of intuition based on data

    I have always used the word “proto-persona” to refer to a concrete step in the process of making “real” personas. (“proto” = “first, beginning, original”) In my experience, they are that gut feeling you get during field research that tells you there seem to be about X number of types of people and they seem to be significantly different on these 4-5 characteristics/dimensions. I like to write those down and keep them around to compare to the patterns I find when I do the more rigorous analysis from transcripts. It’s how I combine top-down (proto-personas) with bottom-up (transcripts/behaviors/motivations) thinking to get a more accurate final product. (Elizabeth Bacon talks about them in a somewhat similar way. Presentation is gone but look in the comments.)

    Proto-personas highlight the importance of intuition in the process of qualitative research. It is fundamental to invention and discovery. It is often the source of the important pieces of “fiction” that end up in a truly powerful persona. Personas based purely on recorded fact without this crucial step of intuition and interpretation are lifeless and usually fail to create empathy or inspire good design. But unlike ad-hoc personas where intuition and gut feelings can easily be untethered to reality, proto-personas are intuition inspired and guided by real data.

    Importantly, proto-personas are not an end or even an interim deliverable. They are a phase in a process and should never be used to make design decisions. I mention them only to highlight how they inform that process. But you have to get to the end. Even though it is hard to wait, you don’t want to bake the bread until it has a chance to rise.

    The point behind all of this conversation is to make us all think long and hard about the tensions in the use of personas and how we can use them carefully so as not to lead our efforts astray. Don’t be afraid to trust your intuition to provide the fiction necessary to bring out the truth. But don’t forget that your intuition must be grounded in good ol’ data.

    There are 3 thoughts on this idea

    1. Elizabeth Bacon

      Thank you so much for the citation! (At least in my browser, the “Death to Personas!” presentation is still displayed there at Slideshare.)

      Very helpful points are made in this post, and thank you for making them. I also want to add some more reference points from the Cooper method where personas were originally developed. At Cooper, we called personas based on intuition & stakeholder knowledge (i.e., second- and third-hand insights) “provisional personas”. This term has some helpful connotations because it implies that we will not rest with such a fuzzy, improvised user description, but rely upon it only provisionally. (Of course I wish that were always true!) I think the term “ad-hoc” has a similar valuable quality, pointing out the sketchy nature of such a persona. There are many risks with half-baked, ad-hoc, provisional personas, as pointed out here.

      Then as Todd also writes, the term “proto-persona” is indeed a step in the process. However, they’re not the INITIAL intuition of what personas will be revealed by user research — that’s what we’d call the “persona hypothesis”. Rather, proto-personas are the initial archetype that emerges from a first cut through the user research data. If you refer to Kim Goodwin’s book, Designing for the Digital Age, you’ll see proto-personas emerge at step 4 (page 256). After interviewee data are mapped to variables, proto-personas arise in the form of patterns that can be identified and explained across the key research dimensions. These emergent clusters of related qualities comprise a “proto-persona” because that data-driven pattern will be modeled further into a fully-fleshed out persona who gains the requisite relevant humanity and detail.

      Hope that helps! Debating semantics can seem tiresome, yet if we all truly shared the same vocabulary, our profession would be better served.

    2. Indi Young

      Agreed. One thing you need to do to gather research for creating personas is recruit participants. To make sure my team thinks as broadly as possible, I perform a little ad-hoc persona type of exercise in-house to define hypothetical “audience segments” to recruit from. Like Jared Spool, I did not want to call them personas, since they are not based on research.

    3. Peter Boersma

      I just read another name for half-baked personas: “user fantasies”. It was reading up on how Agile and UX can work together and found this discussion on (of course) Quora:

      How do UX practices fit with agile development practices?


      There, David Hawdale writes:

      “Although Agile is based on ‘user stories’, these stories can frequently be better described as ‘stories that developers have made up about how they think users act’ – user fantasies.

      Now UX input can fix this one quite easily by validating/creating these stories with users (but definitely not replacing them with ‘stories that UX practitioners have made up about how they think users act’. )”

    Comments are closed.

  • Close
    Team Profile