Katharine Carr
UX Professional

Research Blog

Personas: How to make them – with research or without


Personas are a collection of behaviors or thought patterns among users that are given names and treated as fictional people. They can help clients and product teams think through what users actually need, and start formulating solutions for those needs.

In a perfect scenario where a researcher has ample time and access to users, developing personas would use cross-case analysis: based on data from several user interviews, a researcher would look for recurring behavior patterns and trends (Goodwin, 2009). In the specific process for developing a persona, Goodwin recommends beginning by grouping interviewed users by “role”: the task the user aims to complete. Within the different roles that emerge, demographic and behavioral variables should be noted. This might include “task frequency, mental models, and goals” (Goodwin, 2009, p. 247). Next, the interviewees within roles can be mapped to any common variables that emerge, and placed on a scale “relative to one another, not to what you believe of the population at large” (Goodwin, 2009, p. 254). Analyzing these common variables should lead to some understanding of the users’ underlying goals. At this stage, subtle differences in the user goals should be clarified to avoid oversimplification of the persona (Goodwin, 2009).

The outputs of this process are what could be classified as “behavioral personas,” which are personas grouped by goals, motivations and behaviors – personas focused on what the user does and why (Petersen, 2016). Other approaches to developing personas may be more appropriate depending on the resources available or the needs of a specific project.

Designer and engineer Alan Cooper proposes a more specific, research-heavy approach. These personas are based on quantitative and qualitative data, but in their attempt to be specific, can start to feel unconvincing and frankenstein-like (Petersen, 2016). Another approach to personas involves pulling from the pool of “existing fictional characters from television shows, books and films (e.g. Ebeneezer Scrooge and Bridget Jones)” (Petersen, 2016, p.7). This approach, known as ‘pastiche personas’ from Mark Blythe and Peter Wright, sounds like an interesting exercise but could lead researchers to gloss over some of the important variants among users; in the eagerness to use a fun, pop icon label, we may be tempted to oversimplify and stereotype users. This is in fact, one of the main complaints about personas.

The urge to oversimplify is, of course, human – we like things that fit into our own mental models – but the role of user researcher must remain objective. This is why Goodwin (2009) warns to carefully analyze data, ruling out any behavior patterns that seem obvious but could actually be linked to demographic or other factors. Leaning too much on demographics like cultural background, gender or economic status can warp a persona into a useless mess. This misunderstanding of the goal and basis of personas is common.

A similar misconception is that personas represent the average user. While they do represent commonalities, they should represent a range of behaviors (Goodwin, 2009) and can include special audiences whose concerns have weight, even if they are not the largest segment of the audience. The process of creating a persona – where users are grouped and identified by traits – may also be confused with market segmentations which they are not. Market research usually aims to discover what messages evoke feeling in the audience, rather than uncover how users engage with a product (Goodwin, 2009).

There is some disagreement on how “right” a persona needs to be. To guard against the impact of a faulty persona, they should be created as supplementary or in concert with other design research methods (Petersen, 2016). However, Norman (2008) proposes that it ultimately depends on how the persona will be used, and its weight in the overall research plan and project. Norman promotes the use of “ad-hoc” personas (similar to what Goodwin calls “provisional personas”) to challenge assumptions and build consensus in early stages of a project.

This type of persona can be used when research opportunities are slim. An ad-hoc or provisional persona is low-fidelity and less research-based; instead, they grow out of hypotheses and anecdotes from the product stakeholders and available subject matter experts. The commonalities become a list of characteristics and goals, which in turn become the personas. Even though not bullet-proof from a data standpoint, as Goodwin writes, “a provisional persona still gives you a target, even if its a little off; you’ll never hit anything useful if you don’t aim at something” (Goodwin, 2009, p. 294).

True in personas, and in life.


  1. Goodwin, K. (2009) Designing for the Digital Age. Indianapolis, IN: Wiley Publishing, Inc.

  2. Norman, D. (2008, November 17). Ad-Hoc Personas and Empathetic Focus. Retrieved from https://jnd.org/ad-hoc_personas_empathetic_focus/

  3. Petersen, M. (2016, April 18). The Problem With Personas: How to Make ‘em, Use ‘em and Abuse ‘em. Retrieved from https://blog.prototypr.io/the-problem-with-personas-82eb57802114

Katharine Carr