Empathize
See the problem through the eyes of the people who have it.
Most projects fail not because the solution was bad, but because the team solved the wrong problem. Empathy is the antidote, and it is not the soft part of the work. It is the discipline of putting your assumptions on hold long enough to hear what users actually say, watch what they actually do, and feel what they actually struggle with.
The output of empathy is not a persona deck or a finished journey map. It is a felt sense of someone else's reality that you can defend with evidence. Every later stage of the workflow pays for that sense. Define sharpens it, Ideate stretches it, Test pressure-checks it, and Implement carries it into production.
What you do
Sit in their world
Watch people doing real work in their real context. A screen-share is not enough. Being in the room, or the remote equivalent, surfaces the things they would never think to mention.
Ask open, then sit quiet
The best interview questions invite a story, not a yes or no answer. After you ask, count to three before you fill the silence. The real answer almost always lives in the pause.
Watch the workaround
When someone reaches for a sticky note, a spreadsheet, or a side-channel chat to get part of their job done, you are seeing the most honest signal about what the product is missing.
Hold solutions back
The fastest way to short-circuit empathy is to start solving. Jot the impulse down, save it for Ideate, and keep listening.
What you produce
Interview notes you can quote
Verbatim quotes beat paraphrase. They make the user real in the next conversation, and they protect the team from rewriting history.
Maps of the current state
Journey maps, service blueprints, or simple flow diagrams the whole team can point at and agree on: this is what is happening today.
A list of pains, gains, and workarounds
Three columns that turn empathy into a brief. Pains are the friction. Gains are what the user would love. Workarounds are the gap between them.
AI helps you listen. It does not let you skip listening.
Transcribing, summarizing, and clustering interview notes used to take a meaningful amount of time. AI can make that work lighter. Being with users is still the part that matters most, and that is where the time you save should go.
Use AI to surface patterns across transcripts, suggest follow-up questions, and draft journey maps from raw notes. Do not use it as a substitute for being in the room. The signal you need lives in body language, hesitations, and the things people do not say.
Watch for
Patterns that can make the work look farther along than it is.
Empathy theater
A sticky-note workshop about a user nobody has actually talked to. It looks like empathy, but it produces assumptions. Read the anti-patterns catalog before any kickoff.
Talking to proxies
Interviewing stakeholders about users instead of talking to users directly. The closer you are to the source, the more reliable the signal.
Confirmation listening
Walking in with a thesis and treating every quote as evidence for it. The best researchers actively go looking for things that prove them wrong.
Calling it done too early
One round of interviews is a start, not an answer. The second round catches what the first one missed.
A clear-eyed view of who you are designing for, the friction they face, and the language they use to describe it. Bring quotes, not personas.