Hi Cassandra, I’ve seen lots of recommended personas with varying lengths of details. Some very complicated (pages long) and others really simple like a single line. Are there any templates etc. you use that you could share?
I read about someone advocating for leaving gender, race, and age out of personas entirely because it opens the door to implicit bias. On the other hand, it might be valuable to specifically call out underrepresented groups. What are your thoughts on this?
do you practice BDD and if so are you using personas while defining scenarios? I often think about it and it might help to keep the focus for which customers/users we are implementing features and which behaviour of the system supports different personas.
I don’t personally use templates for personas, mainly because most of the ones I’ve seen aren’t designed for use when testing. For example, a lot of them include the persona’s income. What does that actually tell us about the persona, and their behaviours and feelings towards the product? Not much, in most cases. It might not even be relevant at all after the product has been purchased, or if there’s nothing to do with money in the product you’re testing. For me, it’s very important to tailor resources to the subject and purpose, so templates don’t work for this because they are, by definition, not tailored.
I also find traditional personas too shallow and superficial, only hinting at things that might help us, rather than explicitly stating it. For example, with a persona who has a high income, is the implication that they have a lot of disposable income? Or do they have a lot of financial commitments too, and therefore no disposal income? I find it much more helpful to get straight to the point of why a piece of information might be useful. Templates that I’ve seen don’t do this.
In regard to length, a good heuristic is to think about what you could see yourself actually using on a daily basis. People tend to create far too much documentation that never gets used. Would you realistically refer to a persona that is pages long, or would you be more likely to use a shorter one? Would a one-line persona be enough, or would you need more to go on? When thinking about the level of detail, I’d recommend just enough to guide you well, but not so much that you can’t use your imagination and use one persona for various situations or scenarios. Very much like a test charter.
I think personas for bad actors are a great idea. We’re, unfortunately, seeing a lot of instances where technologists haven’t thought about how their product could be misused or attacked, and they only consider it once it’s out in the open and something bad has already happened. SuperHuman is a great example of this (https://twitter.com/Tweet_Cassandra/status/1146701010250997761). By having - and using - personas for bad actors from the outset, hopefully technologists will think about these things throughout the entire product lifecycle and not just when harm has already been done.
Off the top of my head, there are three instances of this type of persona that I could recommend:
Someone attacking the product itself (security threats)
Someone using the product to attack someone else (exploitation threats)
Someone who just doesn’t follow the product “rules” or workflows - the “but no one would ever do that” type, which is maybe less harmful or sinister than the first two
There’s a lot to unpack there, but I’ll try to keep it short(ish!). Firstly, the time problem / excuse is a common one, and one that might be hard to get around without stepping on any toes. In short, I might say something like, “I understand we’re short on time, but this is really important for our product and our users. Do you mind if I just come up with some new ones and you can let me know if there are any adjustments you’d recommend?” It doesn’t have to take a long time to get something to start off with, or a “model to be wrong”. It’s also interesting that they stopped using those personas, presumably without new ones to replace them. Did they stop using them because the personas weren’t helpful in general, or because they recognised that they were biased and not the right ones to use, or some other reason?
In regard to unconscious bias, I always recommend addressing it head-on. I observe lots of people wanting to hide away from things like bias or diversity because it makes them feel uncomfortable. It makes them feel uncomfortable because it forces them to think about how their behaviours or attitudes might not be morally right, or judged well by others. But the thing about biases is that you have to actively try to overcome them - it won’t happen on its own because that’s the nature of bias. The more we actively try to overcome our biases, the more comfortable we’ll be about these topics because we’ve already starting doing things to improve our behaviour, so that feeling of guilt, shame and / discomfort is lessened, and we can tackle it more and more. Actively think of ways to diversify your personas, and create a set or personas that are different in various ways, instead of different representations of the same group. Research your users and target market; think about their context, backgrounds, etc.
This reminds me of a talk from Eriol Fox, where they talked about how marketing for a product included exclusively white people. The product was built for people in (I believe) Kenya, so when they changed their marketing to include black people, the product uptake increased greatly because users could actually relate for the first time. You can watch the talk here: https://think-about.io/events/2019/speakies/eriol_fox.html
That’s a really interesting and tricky question. I think the point in using personas is to challenge ourselves to think about different people who might use our product, and to always keep our users in mind. Do biased personas that are largely the same help us to think about different people? Probably not. Could we each think of different users ourselves and keep them in mind in our daily work without a documented reminder? Maybe some could, but I think most wouldn’t. I think both options are dangerous. What do you think?
Ideally, I would recommend having personas from the very beginning. However, it’s not always that simple. Traditionally, personas are used for marketing, but marketing personas aren’t always useful for other purposes like development or testing, as different information and aspects are important for each. That doesn’t mean the personas have to be completely different, but they should be targeted for a specific purpose so they’re more useful and likely to actually be used.
If I was in charge of everything(!) from the start, I’d start off with less detailed, simplified personas for different purposes (think of one person having a CV, a Twitter bio and a dating profile - different things are included in each) and then develop and adapt them as things progress and more user research is done. I think it would be silly to keep personas the same forever, as that isn’t realistic, especially as you gather more information about how people use your product, and who they are. Your target market might change too, or you think of more important things to include in personas. Do what makes sense, and is actually useful.
I touched upon this in another answer, but I’ll try to be more specific here. The idea that implicit bias is avoided by not talking about or actively striving towards diversity is fundamentally flawed. As humans, our brains are wired to fill in the blanks, based on our own un/conscious biases and mental models. If we’re told to imagine a CEO, most people will think of an older, white man. If we’re shown a picture of a young, Indian woman and told that they are a CEO, I think most people would be a little surprised, whether they admit it or not, because it doesn’t fit into the idea of a CEO that they un/consciously have in their heads. If we completely omit gender, race, and age from our documented personas, our minds will still fill in the blanks for us and our implicit bias will still come through in the undocumented parts of the personas. The only difference is that we’ll do it individually, instead of sharing the biases of whoever created the personas.
My advice: include protected characteristics in personas and make an active effort to diversify. Don’t just include things that you can assume from looking either. Consider invisible factors too, like mental health, religion and chronic pain. Ask multiple, diverse people to create and review them to help counteract the bias of a single person or group. If you can’t find diverse people in your company, recognise that as its own issue and tackle that too.
Traditionally, this comes down to user research, which is usually conducted by a UX team. When a product is new, I imagine that marketing teams base personas on the target market and who they’d like to use the product, which, of course, doesn’t always turn out to be the reality. When it comes to using personas for testing, I question how useful the information in traditional or marketing personas is, and in my talk at TestBash Germany in September, I’ll talk about moving away from traditional demographics and focussing on mental states instead - something that is fluid, changes frequently, and is largely not considered at all in traditional personas.
As a side note, I wonder to what extent we can really validate our personas based on real user data, without violating users’ privacy or having skewed datasets because of the type(s) of people who are more likely to agree to sharing their data for those (or other) purposes.
I’m not sure what you mean by “product personas” (whether they’re used for marketing, thinking about role / permission types, etc.) but in general, I think personas should be fit for purpose and where the purpose is different, the information in the personas should be different too. As I said in another answer, this doesn’t mean that the personas themselves are different for each purpose, but that the information included is different (e.g., CV vs Twitter bio vs dating profile) What is considered useful and relevant differs, based on the target audience and intended purpose.
What to include in a testing persona can differ largely based on the product. I’d always recommend keeping the product in mind and thinking of things that make sense in context. Some things you might want to think about include: technical proficiency, gender (not just binary), ableness, mental state, travel patterns, personality. It all depends on context and what interesting ideas or scenarios could come from each aspect.
Just as there are lots of ways to test, there are lot of ways to use personas in testing. Some examples include:
Keeping personas in mind during three amigos meetings and design reviews, and thinking of how features and / or designs could cause a problem for some users
Creating test scenarios based on specific personas and their behaviours
Performing testing tours as each of the personas
Do you have any other ideas of how to use testing personas?
I too often see “edge case” used as an excuse for exclusion, and agree that it’s dangerous. It’s also lazy and dismissive. I believe that we should strive for diversity in our personas, and when we have diversity in personas, it helps us to have diversity in our scenarios and ideas about how a product could (or not) be used too.
Of course, we don’t have all the time and money to build everything we want to, but that doesn’t mean that we should ignore stress cases and pretend they don’t exist. Personas are there to remind us of our users, and if we are reminded that we are excluding some people, then we should be reminded. If we choose not to accommodate a stress case at a given point in time, we should be forced to pinpoint a better reason for that decision than “edge case”, and understand the impact that decision could have on users.