Personas From Evidence
Build a persona that comes from your research, not from stereotypes.
What you will learn in this lesson
You will build one persona from your research. You will also learn how personas can go wrong.
What a persona is
A persona is a short profile of a type of person you are designing for. It is built from real patterns in your research. It helps your team remember who the design is for.
A persona is not one real person. It is a pattern from several people. Never use a real participant's name or photo.
What goes in it
Keep it short. Focus on behavior and goals, not looks.
| Part | Example: pantry visitor persona |
|---|---|
| Name and one line | "Marisol, a working parent who visits the pantry twice a month" |
| Goal | Get food for her family without missing work |
| What she does now | Comes on her lunch break, takes the bus |
| Frustrations | Got turned away once for missing a utility bill |
| Tech | Android phone with a prepaid plan, uses WhatsApp and Facebook |
| A real quote (no name) | "If I'd known, I would have brought it." |
Personas can go wrong
Personas can turn into stereotypes. That is a real risk, and it causes harm.
- Don't guess traits. If no one in your research said it, leave it out.
- Skip details that don't change the design. Age, race, and looks rarely matter to the task. Adding them can bring in bias.
- Include access needs when they come from research. "Reads small text with difficulty" or "uses a screen reader" changes the design. Put those in.
- Watch who is missing. If every persona is a confident smartphone user, you will design out everyone else.
Use AI carefully
AI tools will happily invent personas with made-up details. That is the problem. Those details come from patterns online, which often carry stereotypes. If you use AI, give it only your real, anonymous patterns. Then delete anything it added that your research does not support.
Prompt Lab
Turn your real research patterns into a draft persona. Then remove anything the tool invented.
Your prompt
Act as a UX researcher who writes personas only from evidence provided and flags anything not supported. Format: A short persona with: one-line description, goal, current behavior, frustrations, tech access, and a list of any details you guessed