In defence of personas

Personas are a tool that are used a lot in agile development processes. They are authentic representations of different types of users. Through this representation they communicate information about how user behaviour may vary when the site or service is viewed from different circumstances.

Personas have been interpreted as tools that encourage developers to ad lib on top of scientific data. This is not how I think of personas. I have not always been a supporter of the persona and in this blog I will explain my journey from resisting to advocating.

I entered the world of user research as an anthropologist. It was a baptism of fire, moving from detailed prolonged ethnographic approach that is in search of representational truth, to bootstrapped fast-paced gathering and application of insight that makes essential services easier to navigate for the people who use them. And more than that, to ensure that there’s an equality of accessibility for everyone looking to use the site or service.

At the start of this process I felt that personas were a massive injustice. They skipped all of the rich detail that translates data to understanding. They encouraged developers to think of people in isolation and in segments. Often one or two personas surfaced to the top as the ‘main’ users with everyone else demoted to ‘edge case’ status. Not only did this introduce bias into the decision making process but it reduced the quality of research and insight to stereotypes. Very dangerous for public sector services.

My gut reaction: something is broken, let’s fix it.

After a bit of subtle investigation, it did not take long to identify breaks in the process that reduced the personas to stereotypes.

  1. Personas were estimations of user activity based on blunt data tools like analytics.
  2. Developers were isolated from the research process.

After a bit of investigation and experimentation, we applied two quick solutions.

  1. Apply qualitative research to the personas to make them representative of how people act.
  2. Involve more developers in more research.

The fix

In anthropology there is a controversial ethnographic tool that is used when it’s difficult to guarantee that participation in research won’t have repercussions for the individual. For example, ethnographic research with criminal gangs. In this case, simply changing the name of the individuals is not enough to protect their identity as they might be identified by events or actions during the study: a hard sell for the anthropologist looking to learn how and why criminal gangs are formed.

One solution some anthropologists have tested is to compile events from multiple individuals into a representative character. There are some golden rules:

  1. The individuals that feed into a representative character are always of the same social status within similar communities, with similar relationships and activities.
  2. The conversations, events and actions are always real conversations, events and actions that have been observed.
  3. A series of interrelated events is never broken

By representing information this way you can protect individual identities, whilst still describing authentic relationships and events. You can convey how and why they work, and identify pivotal moments without breaking confidentiality.

This approach can be risky. You’re no longer reporting exactly what someone said to you, which introduces more room for bias and error, especially if you don’t have much research to work from. However with these risks come some significant benefits. It means we can:

  1. Focus on what motivates a group of people, how they problem solve, the tools and actions they use to get things done.
  2. Convey a huge wealth of information quickly and in a relatable format.
  3. Allow research findings to be widely shared and used, even in situations where sensitive personal data or circumstances are involved.

How to make personas work harder

The key to making personas representative and effective is research and data.  Having a good understanding on the circumstances people are in, the dependencies they have and how they relate to a site, helps us to make sure that we’re building user focused services as a whole.

Further Reading:

Argument against personas:

Ethnography that uses ‘personas’:

Persona template and case studies: