uxGPT: Creating personas with prompts for better questions using ChatGPT and uxGPT Personas
Patrick Neeman
UX Leader at Workday. Author of uxGPT: Mastering AI Assistants for UX Designers and Product Managers. Working on great things with Gen AI since November 2022. Ex-Microsoft. Opinions are mine.
Designers: You can’t just chat with an AI avatar representing an average of your customers. Also designers: This is Jane, a persona I made based on how I imagine our customers. Let’s make all of our decisions based on Jane.
There’s a lot of truth in this.
But back to why we are here.
When I walk into a new place, the first activity I do is gather information to create personas?—?it could be user interviews, usage data, or whatever?—?because there has to be some baseline of the users.
I’ve been in the legal technology space for a while, so I have a pretty good idea of who the users are. In other places, I would have to start from scratch and use my lightweight personas approach using data as much as I can and use my personas template for the artifact.
Those places usually have personas but are a wonderful mess that are more fiction than research. It always surprises me how often previous personas seem more made up and not based on data.
Based on that tweet, the fancy tape recorder of ChatGPT frankly might be a better baseline than what most organizations have, making this one of the most maligned conversations in how user experience is being affected by AI.
The goal of this exploration is not final answers, but to quote Ha Phan, asking better, more informed questions.
ChatGPT can generate better information to ask better questions because of the vast amount of information used to build the models (admittedly without the permission of some users), then we could.
With a trust-and-verify model, you can generate better user research questions.
However, I’m not here to start a controversy?—?I’m here to teach with available tools. Let’s get started.
Start wide
First, let’s define what a good persona should have because they are one of the most misused artifacts designers and product managers use. Alan Cooper, the inventor of personas, has a great article about the journey.
A good persona should focus on goals, motivations, and behaviors because these elements provide a clear, actionable understanding of the user.
By concentrating on these aspects, personas are powerful tools for aligning stakeholders with user needs, fostering empathy, and ensuring that solutions are both relevant and effective. This focus eliminates unnecessary details that could dilute the insights, keeping the persona practical and directly applicable.
I’m going to start with the most basic of prompts, using the application customer relationship management as with the other articles in this series.
Prompt
Create a persona for a Sales Business Development Representative using a customer relationship management application. Frame the persona with goals, motivations, and behaviors.
Add structure
Let’s get rid of the fluff that Chat GPT added.
A good persona skips demographic details and focuses on background and attitudinal information because demographics often fail to reveal true user needs (age, really?). Instead, understanding a user’s background provides context, shedding light on their experiences and challenges.
Attitudinal information captures their mindset and perceptions, which are crucial for tailoring the user experience. By honing in on these aspects, we gain actionable insights into how users think and feel, driving design decisions that resonate more deeply.
This approach keeps personas relevant and focused, ensuring we design solutions that truly address user pain points and aspirations.
Prompt
Create a persona for a Sales Business Development Representative using a customer relationship management application. Frame the persona with goals, motivations, and behaviors. Include attitudinal information like pain points and background information. Do not include demographic information about this persona.
Add the detailed?task
Now let’s include research that almost never happens because it’s too much work to interview users this way: the task they are performing as a persona.
Organizations use personas in the abstract because they are a lot of work to create, so they speak of them. I’ve been in many meetings where we speak of the persona but don’t actually interview users that match the persona. If we do, it’s usually after the fact during usability testing.
When we specify the exact tasks users need to accomplish or the features they interact with, we create a direct line to practical improvements. This detail helps prioritize what matters most to users, guiding the development team to focus on functionality that genuinely enhances the user experience.
Personas are too abstract without including this detail. Including detailed tasks and features ensures our designs are grounded in real user needs and behaviors.
A task-based approach also turns the persona exercise into an evergreen activity, refining them based on real research versus at the beginning of a?project.
Prompt
Create a persona for a Sales Business Development Representative using a customer relationship management application, specifically managing accounts. Frame the persona with goals, motivations, and behaviors. Include attitudinal information like pain points and background information. Do not include demographic information about this persona.
Trust and verify with user research questions
Asking user research questions to validate a persona ensures it accurately reflects real users. This is crucial for personas because they must match someone in reality. This validation process roots the persona in actual data, making it a reliable tool for design decisions.
Skipping this step risks creating a persona based on biases or guesswork, which can lead to misaligned solutions.
User research is the compass that keeps our personas?—?and our designs?—?true to the real user experience.
We can easily do this with GPT by adding to the prompt to ask user research questions, as we discussed in the last article.
Prompt
Create a persona for a Sales Business Development Representative using a customer relationship management application, specifically managing accounts. Frame the persona with goals, motivations, and behaviors. Include attitudinal information like pain points and background information. Do not include demographic information about this persona. Generate user research questions that would validate this persona as correct.
Others to?follow
Try out this Custom GPT?—?uxGPT?Personas
Don’t want to do the work yourself? Not a problem. I’ve done a lot of the legwork for you.
Usability Pioneer | UXtigers.com | ex ????????
8 个月Great. I would experiment with modifying the last line in your last prompt to “Generate user research questions that would validate whether this persona is correct or needs changes.” (I.e., don’t assume it’s correct. Maybe in fact assume that changes will be needed, because user research always uncovers something new).
Product design and UX leader
8 个月I like the idea of using generative techniques to come up with good questions, in this case to test the ideas embedded within a persona description. I wonder in what ways the prompt might push the critique, like: - generate a list of assumptions about why the persona has these motivations, and what types of research or analysis would be needed to effectively evaluate whether these assumptions are true or false - generate a list of behaviors that the persona can be observed doing, and what types of research or analysis would be needed to observe this effectively - generate a list of statements in form of "Hypothesis: because I [motivation], then I [behavior]" that expresses the assumed cause and effect between each of the persona's motivations and each of the behaviors that stem from each motivation. For each statement, list research methods through which a researcher could effectively observe the cause and effect