Demographic
Synonyms: demographic data, demographic variables, user demographics, audience demographics, population characteristics
Definition
Use cases
- Skewed sample: A banking app tests with 8 staff, all under 35. It ships with 12px labels older customers can't read.
- Stereotype persona: "Sarah, 42, mother of 2" gets a stock photo and a yoga habit. None of it came from research.
- Intrusive form: Signup asks for gender and income on screen 1 with no reason given. People quit or answer at random.
- Hidden subgroup: Checkout errors look normal in aggregate, so a group failing at 3 times the average goes unnoticed.
How it's used in practice
- Set recruiting quotas. For example, at least 2 of 8 participants aged 60+.
- Ask only what changes a decision. If an answer wouldn't change anything, cut the question. Recruiting panels already store age and location.
- Put questions last and make them optional. Say why you're asking, and offer "prefer to self-describe" in place of "Other."
- Segment, then investigate. Split analytics by age or device to find who struggles. Then interview that group.
Challenges & limitations
- Gaps without causes: NN/g found users 65+ were 43% slower on websites than users aged 21 to 55. The number doesn't say whether text size or tap targets are to blame.
- Sensitive data: Racial or ethnic origin and health data are special-category data under GDPR. Collect less, explain why you need it, and protect or anonymize it.
- Small samples: Splitting 8 usability participants by age leaves groups of 1 or 2.
Free resources
- NN/g: Why and How to Use Demographics in UX — when demographic questions earn their place, plus wording and ordering guidelines for screeners and surveys
- NN/g: Handling Sensitive Questions in Surveys and Screeners — why early demographic questions cause dropoffs and how to build respondent trust first
- NN/g: Personas vs. Jobs-to-Be-Done — why demographics alone are weak inputs for product decisions and what a full persona includes
- NN/g: 3 Persona Types — proto, qualitative, and statistical personas, and why demographic-only personas lack context

