if your users are *really*
just like you (e.g. 37 Signals) or you’ve designed for this user group successfully before and you already know them, you may not need to research. if you ﬁnd yourself using stereotypes and/or cliches... or using family members as examples, you probably need to research.
there are lots of different
ways to research quantitative (e.g. surveys.) qualitative (e.g. interviews, card sort) ethnographic (‘user stalking’) longitudinal (continues over time) contextual (in the context of use) lab based (semi contextual if possible) ... and more
your best standard research approach
will be a series of 45min interviews with 6-8 participants. ... but think about: - what you’re trying to learn - how you could make it as contextual as possible ...don’t be afraid of getting a bit creative
> introduce the session (esp.
inform and ask permission re: recording) > good rapport = good research (take time to make it clear you care about their opinion) > don’t tell them you’re the designer (they’ll not want to hurt your feelings, they want to please) > it’s not a test (if anything is being tested, it’s the design - NOT the participant) > only *their* opinion counts (we don’t care about what other people might think or do)
> deﬁne your research questions
> plan & practice the ‘script’ > start as wide as possible, narrow slowly > keep it contextual, not speculative > show, don’t tell (observation = good) (show me how you do that?) > uncover mental models (if you clicked there, what do you think would happen?) > ask OPEN questions (ask questions to get them talking NOT yes/no answers) > take care not to lead (you can have leading questions AND leading structures) > never leave them hanging (if they’re struggling, save them)