Our understanding of psychology is growing and changing constantly. In universities and research centres, the fundamental concept of what a human mind is, and how it works, is evolving rapidly. The everyday world of work, and products and our daily lives lags behind in this understanding however, and as a result a lot of the little ‘templates’ of understanding that we work from to make things easier for ourselves end up being outdated, and incorrect.
I believe that for many product design situations, the concept of ‘personas’ is one such outdated idea. We can’t argue with the fact that is has some value, but might there be a similar but importantly different model that yields a lot more value, and that is significantly more accurate?
First think about the value that persona development offers, and some of the assumptions it is based on. Personas are small ‘pen portraits’ of different stereotyped individuals, that we are expecting to use or buy whatever it is we are making. Each individual persona usually includes various characteristics, such as ‘motivations’, ‘technical abilities’, ‘skills’, ‘goals’, ‘behaviour patterns’, etc. The value then is of course to inform a user-centred design and development process, whereby we always make sure that we are designing something that meets the needs, abilities and limitations of the people we are designing this thing for. Usually we are not designing something for everybody, so working with personas should help us ensure we are making design decisions that will specifically benefit our expected users.
This is based on a few assumptions though. There are assumptions firstly about the accuracy of the personas (whether the actual personas you have really do match up realistically to potential customers and users) but I won’t be dealing with those here. I’m more interested in another assumption – that our personalities and behaviours are consistent over time and environments/contexts. You wouldn’t usually think of this as being one of the assumptions of using personas, but that’s only because it is so deeply baked into the idea. Imagine you create a persona of ‘Paul’, who is a highly skilled web developer, works from home, has a dog called Lucy, loves trail running, has high attention to detail, values physical exercise and time with family above money or status…and so on. Well, in the everyday use of this persona, this assumes that Paul will consistently have high attention to detail, will always value exercise and family above money or status, etc. A lot of research suggests, or even proves that this isn’t a very good assumption to make.
In fact, it appears that we can behave quite differently given different social and physical environments. Not only that, but also we think very differently as well, when presented with different kinds of information, or the same information presented in different ways, or when in different emotional contexts.
So for example, Paul might generally be described as someone with high attention to detail, but we now know that we all only have what can be thought of as a limited supply of cognitive resources to work with during the course of a day (for example), and the more depleted they are, the less able we are to be highly detail oriented, or indeed to perform any highly effortful cognitive task. If the product you are designing is one that will generally be used by Paul at the end of a work day, he might actually exhibit less attention to detail than someone with generally low attention to detail, who’s just had their morning coffee.
Another, perhaps more telling example: I recently read Thinking, Fast and Slow by Daniel Kahneman, the legendary Decision Scientist/Psychologist/Economist. The main thesis of the book is that all of us operate two systems of cognitino and decision making, which Kahneman names, not very revealingly, ‘System 1’ and ‘System 2’. System 1 is characterised as ‘fast, automatic, frequent, emotional, stereotypic, subconscious’, while System 2 is characterised as ‘slow, effortful, infrequent, logical, calculating, conscious’. We all use both systems, but what is important is that we use a particular system based in part on the way information is presented to us, and also in part on our environment and state of mind.
The book also deals a lot with resulting biases and common errors of human information processing. Here’s the kicker as far as I am concerned: The different ways the same person will think and behave can actually, and often, end up being more different than two separate, traditional personas. I’ll put that another way. Every single real person regularly thinks and behaves in ways that contradict any kind of definable, consistent characteristic. Think of what you might include in a persona of a senior fund manager. Would you ever include that that person regularly makes errors of probability reasoning that result in their choices being less accurate than pure chance, even though they can happily pass an advanced test of probability reasoning? Well, that happens, and many other things like it.
If this is all seeming a bit impractical, then stick with me! This doesn’t only apply to mathematical kinds of thinking, it applies to ordinary, everyday information processing and decision making done by ordinary people. My suggestion for an alternative to Personas, is the development of Contexts & Modes. Let’s imagine that we are designing a new project management software tool, that should help people better understand how to shift schedules and timelines around, based on how different bits of a big project are going.
It is important that we think about Contexts before Modes. Contexts can be thought of as the context and environment we expect our users to find themselves in, when they are mostly going to be using the thing we are designing. This is not just their physical environment by any means, but perhaps more importantly their social and emotional context, as well as their physical one.
So for a trivial practical example, we might define two contexts for the use of our product. One is ‘Moments of downtime at work’, characterised by being rushed, amongst other people, cognitively depleted and in a relatively competitive environment, where it is important to appear successful, decisive and capable. The other is (sadly) ‘Sunday evenings at home’, characterised by being with family, not cognitively depleted, in a non-competitive environment, but obligated to engage in ‘social’ conversation at any time.
OK, now that we have defined these Contexts, let’s consider how they might shape the modes that our users operate in.
So these modes can be thought of as modes of thought and behaviour, that any person can be in, regardless of their supposed characteristics and personality traits.
Considering the two Contexts just described, the first one could result in, for example, a mode of ‘Desiring social acceptance and increased use of System 1’. What does this mean? Well, since the person is at work, and driven to appear capable, wanting to be accepted and respected etc, he will likely be more susceptible to suffer from the ‘sunk cost’ bias. This is where we tend to keep investing in something, even it appears like a sinking ship, just because we have invested so much already. It is thought in large part to be an attempt (ironically) to avoid feelings of regret. Secondly, the increased use of System 1 is a result of the cognitive depletion and busy work environment. This means our user will be less likely to spend a lot of conscious, cognitive effort working through problems, and as a result be more at risk of cognitive biases and rely more on heuristics, like the ‘availability heuristic’. This is when we think something is more important than it necessarily is, just because it is easy to think of examples of it. An example is how likely we think it is to die in an air crash. It is easy to bring to mind examples of plane crashes, because of media exposure, so we tend to overestimate the likelihood of it happening, when thinking about is quickly.
In contrast, in the second context our user could slip into a mode of ‘Relaxed focus and increased use of System 1’. The social environment has changed drastically, and our user is no longer in a context in which a primary drive is to appear capable and successful. He is not cognitively depleted either, but we have hypothesised that because he is engaging in light social conversation throughout the usage period, he is still mostly going to be relying on System 1 type processing, rather then the more effortful and energy consuming System 2.
So what’s the end point here? Well, we need to present the information in our software in a way that takes the active modes of our users into account. Our user (we can keep using the example of ‘Paul’) would be described usually as a software developer/project manager, with good mathematical and logical reasoning abilities, and high attention to detail. However, based on the modes we have calculated, he is actually going to be in a frame of mind when using the software that will result in him neglecting important information about the long term benefits of a project, unable to process probabilities well, and likely to make a decision based on what most easily comes to mind rather than a more detail oriented approach of considering various options and weighing them up.
Since our software is supposed to help Paul make better decisions than he could on his own, and some of these decisions are about shifting emphasis and timelines for different projects based on how they are doing, this is all pertinent information. If we relied on our traditional persona, we could present some detailed information to Paul in some well designed tables, detailing the amount invested in each project so far, and the expected yield, and progress so far etc. But given all this great data, and what we actually know about Psychology, and Paul’s real world Contexts and Modes, this would be a poor design decision. Instead we should present less information to Paul, perhaps even hiding any figures relating to investment, and only presenting a visual indication of which projects are yielding the most value, and how they should be re-scheduled to make them work better. In this way we protect Paul from being a victim of the sunk-cost bias, and do something that truly great tools should all do – extend the ability of the human mind or body, to allow us to do something better than we can do on our own.
Now I’m not saying throw your personas out of the window – they are better than having nothing. However they can actually be misleading and damaging, and we need to start thinking more clearly and accurately about how people really are. Paying lip service to some UX process or other, and creating pretty personas means nothing. Being a good designer means more than understanding the people that will be using what you’re designing – it means understanding who they will be at the very time and place they are using it, because that could be someone else entirely.