Review
Klein writes for people who have already been told to do user research and are stuck on the practical question of how, with a real team and a real deadline.
The structure is a loop: understand the users you have, decide what to change, build the smallest version, measure whether it worked, repeat. That is not novel. What makes the book useful is that each stage comes with a specific exercise rather than an exhortation, and the exercises are sized for a team that also has to ship something this month.
The section on personas is the one I'd defend hardest. She's blunt that most personas are decoration built from demographics, and rebuilds them around behaviour and motivation instead. A persona that tells you someone is 34 and lives in a city has told you nothing; one that tells you what they're trying to get done and what stops them changes a backlog.
It's aimed squarely at the earlier part of a career and it doesn't pretend otherwise. If you've been running discovery for years you'll skim large parts. If you're setting up the habit for the first time, it's better company than most of the more famous books.
Key Takeaways
The parts worth keeping:
Personas built on behaviour, not demographics
- Age, income and location rarely predict how someone uses a product. What they're trying to accomplish and what currently blocks them does.
- A persona should be specific enough to rule a feature out. If it can justify anything, it is decoration.
- Build them from conversations you actually had, not from a workshop where the team invented a plausible person.
Change one thing and know why
- Start from a stated belief about what is wrong and what you expect fixing it to do. Without that, whatever happens next will be explained after the fact.
- Bundling several changes into one release makes the result uninterpretable, which is usually how teams end up with confident theories and no evidence.
Qualitative and quantitative answer different questions
| Method | Good for | Bad for |
|---|---|---|
| Analytics | How many, how often, where they drop | Why any of it happens |
| Interviews and observation | Why, and what people are really doing | How common it is |
- Using one to answer the other's question is the most common research mistake. Numbers cannot explain motivation and five conversations cannot size a market.
Habits over projects
- Research done as a project ends and its findings decay. Research done on a cadence compounds.
- A small amount every week beats a large study every quarter, because the small version influences decisions while they are still being made.