Creating user personas for open data

Use these tips as a guide to compile ethnographic observations into a format that allows designers to relate to users at a glance

Step 1: User Discovery

Generating effective personas begins with research, which requires going out and talking to community members. The point of user discovery is to elicit stories that can emerge in patterns. Scheduling and conducting interviews accounts for most of the time involved in creating personas. Two to four weeks is a good window for lightweight user discovery sprints.

1A Prepare Your Team

Make sure you operate within a focus area. Narrow the scope of the design problem before starting user discovery by considering city or local objectives. Making preliminary personas quickly from your team’s present knowledge can help to gain a sense of what kinds of attributes of user categories are pertinent to local objectives. It is important to think about what characteristics you want to compare across different segments of the population. As a team, think about preliminary categories of users, but be ready to adjust them later. Who are the open data users? What are people going to get out of open data? What public information do they need, and what is their role in their community with regard to open data? Test assumptions by trying to find information on the portal as if you were one of the personas.

  • Identify a team of 3–4 people to work on all stages of the process of creating personas for open data. The final products are composites, and so personas are best assembled by team members that bring different expertise across City departments or disciplines.
  • Identify key lines of inquiry about a problem and its context. Key lines of inquiry might be motivated by a citywide or agency objective, such as ending homelessness.

  • Run a preliminary personas workshop.

1B Conduct Key Informant Interviews (KIIs)

Key Informants may include local community leaders, relevant community organizations, civic tech groups, or relevant city staff in other departments. Often there are small groups with loud voices that are disproportionately represented in how cities make decisions. At this stage, cast a wide net to learn who is out there, what their problems are, and what support they are in need of. Allow key informants to tell you about the open data landscape. KIIs should comprise of mostly if not entirely open ended questions.

  • Conduct ~5 KIIs for gaining a wide initial understanding of the open data user landscape. Ask Key Informants for referrals to interviewees.

1C Conduct User Interviews

Ask open-ended questions that don’t have short answers, i.e. “Why do you use open data?”, “What is your experience with city services in general?” Follow up with “How?” and “Why?” Staying aware of Key lines of inquiry may surface goals, values in the community, technical skills, non-technical skills. Don’t follow the question guide precisely; allow the interviewee to guide the conversation and tell his or her story.

  • Prepare a question guide of questions you might ask. This can be done as a full team or in smaller groups.
  • Determine ~5 key questions that you want to know about every interviewee.
  • Prepare ~5 additional questions specific to the context of the interview.

Rigorously distinguishing between observation (what is empirically verifiable) and interpretation (the story we tell ourselves) can help to avoid biased stereotyping. Use a template for interview notes that distinguishes between facts and observations. Recall that an observation is something that anyone can verify empirically. In a user discovery interview, an observation may be about body language, tone of voice, or reporting a quote. An interpretation is our explanation of why something happens or what something is that we can’t observe, for instance “the interviewee is harboring a silent distrust of open data”. More interpretations can always be developed at a later time, whereas more observations cannot be collected without conducting more interviews, so try to record many observations.

  • Conduct ~20 interviews. More is even better.

Make sure participants know that they are valuable to the research, especially if they are not current users. Think about what questions might acknowledge their relevance to the research, i.e.  ”What are other times that you have solved a problem, and how?”. Keep in mind the goal of the research while conducting user discovery. What would people like to solve? What are the current and potential use cases?

Step 2: Synthesis

The most creative aspect of generating personas is taking information from many observations, and distilling it into one fictional, yet realistic and research-informed persona. Synthesis can be completed during one meeting. Reconvene as a full team to synthesize interview notes into patterns.

2A Meeting Preparation

  • Set the agenda for the synthesis meeting before starting. As a group you should articulate what you want to use the personas for. Return to your initial questions or design problems. Share and display the key lines of inquiry. Explicitly agree about what characteristics you want to compare to define and narrow the scope. Each team member should prepare for the meeting by reviewing interview notes, making comments, and beginning to look for patterns.

2B Use affinity mapping to identify segmentations

With key lines of inquiry in mind, try to solidify approximately five categories of interest across all personas. Experiment with clustering post-it notes by theme–there are many possible ways of grouping observations, so don’t stress about finding the “right” category. If you can explain a cluster of observations to others in your group, stick with it. Individual interviewees should not be recognizable in the process affinity mapping. This allows for abstract patterns to emerge. Personas should not present web or survey analytics.

  • Write down observations on blue post its
  • Write down interpretations on yellow post its
  • Write down patterns on green post its
  • Make sure the team is more or less on the same page about observations and interpretations.
  • Mark two axes of interest on the wall. The axes, ranging from low to high, should relate to key lines of inquiry. For example axes might be: number of open data use instances, technological ability, proximity to government, quality of community impact, number connections in the community, access to other data sets.
    • Merging themes and patterns is easier when there is a framework for organizing data. This will give the team a sense of what to look for and pull out for the personas.
    • The segmentation should allow you to compare characteristics across different segments of users
  • Look for patterns and themes within clusters. Which observations resonate together?

Step 3: Persona Generation

Finalize the synthesis process by compiling the results of brainstorming and affinity mapping. Use existing personas or templates as a frame of reference, or create your own design. Small anecdotes can help round out a realistic character while also signaling useful information. For example, portraying New York City’s busy bystander with headphones conveys a sense that she is preoccupied, while allowing us to think of her as a real person.

Distill segmentation information into a composite individual persona

Persona characteristics should be anonymous, yet specific. The appropriate level of granularity will depend on what you want to use the personas for. A citywide redesign of the open data portal will entail more general characteristics than those of a set of personas developed for addressing a specific issue such as homelessness or contracting.

  • Prioritize roles, motivations and needs over occupation titles or group affiliations, and strip back demographic detail to a minimum. Think creatively about how to present the characteristics.
  • Think creatively about titles that describe roles (not just occupation). A title should convey what you want to compare across the open data economy?
  • Include capabilities
  • Include incentives
  • Include personal anecdotes
  • Include a picture
  • Once you have a title and some characteristics for each persona, think about their Network and their position in the open data economy
  • Look at models of existing templates
  • Five or six is a good number of personas. Two is probably too little, ten is more than plenty.