2learn

Empathise, define, ideate

What it is

The first half of the design-thinking loop moves from watching people, to a single agreed problem statement, to a pile of candidate answers. Empathise is the study of what people do and why. Define compresses that study into one sentence worth solving. Ideate widens the field again before anything is built.

Why it matters

Jumping straight to solutions feels productive and is the most expensive habit in design. A team that has not watched the work guesses at the user’s language, the constraints and the workarounds already in place, then builds for a user who does not exist. Writing the problem down also settles arguments that discussion cannot: two people who disagree about how to build something usually disagree about what the problem is, and one sentence makes that disagreement visible while it is still cheap to fix.

Core ideas

Empathise means watching, not surveying

Sit with people doing the real task in the real place. Record what they touch, what they say while doing it, and where they pause. Opinions about a feature are the least useful thing you can collect; behaviour is the most.

14:02  opens the order screen, scrolls past search, uses the "recent" tab
14:04  mutters "where is it" — repeats the same scroll
14:07  gives up, phones the customer to ask for the order number

Three lines like that say more than a survey score, and they hand you the user’s own words for the define stage.

Define is one sentence in the user’s words

Compress the evidence into a statement naming the person, the goal and the obstacle. It must be specific enough to rule options out.

A warehouse picker halfway down an aisle needs the next line item without
walking back to a terminal, but the pick list exists only on the terminal.

If the statement would fit any product in the category, it is too broad to test.

Ideate for quantity, then filter against criteria

Generate as many answers as a group can inside a fixed, short window and forbid criticism while generating — a judgement spoken at minute two removes the idea that would have arrived at minute twelve. Filter afterwards against the problem statement and a named constraint such as cost or build time.

Rules posted on the wall:
1. No "yes, but" until the timer stops.
2. Build on the previous idea before offering your own.
3. Volume is the score — aim for 30, expect to keep 2.

Diverge and converge as labelled modes

Say which mode the room is in. Divergence adds options; convergence removes them against criteria. Unproductive workshops are usually stuck in one mode without noticing: generating long after the decision was reachable, or criticising through the only ten minutes that could have produced a new option.

Practice

Write a define statement from raw notes

Collect fifteen observation lines from a task you can watch for an hour — a queue, a checkout, a support call. Write the problem statement in the user’s words, then have someone who was not present read it back and say what it rules out.

Done when: the reader can name at least two solutions your statement excludes.

Run a twenty-minute ideation

Take your statement to a group of four and run one timed divergence round under the three rules above, then a fifteen-minute convergence against cost and effort.

Done when: you have thirty numbered ideas and a shortlist of two with the criterion written beside each.

Common mistakes

  • Interviewing instead of observing — people report what they believe they do, which drifts from what they do; fix: watch the task performed and write down the steps as they happen.
  • A problem statement that names the solution — “users need a dashboard” has already chosen the answer and cannot be tested; fix: state the goal and the obstacle instead.
  • Criticising during divergence — the room converges on the first safe idea and calls it agreement; fix: separate the timed generation round from the filtering round.
  • Stopping at the first workable idea — the obvious answer is rarely wrong but is often mediocre; fix: keep generating past it and compare at the end.
  • A problem statement with no evidence behind it — it is a guess wearing a process; fix: trace every claim back to a note from the empathise stage.

Go deeper

Worked examples

Analytics cookies are set only if you allow them. Nothing is loaded, and nothing is sent, until you choose.