The STAR method

What the question is testing

Behavioral questions ask for evidence, not opinions. "Tell me about a time" means one specific episode: a project, a decision, a week that went wrong. The interviewer is checking that you can recall real situations, that you were the one acting in them, and that you can explain clearly what happened and why it mattered. A general answer ("I usually...") gives them nothing to score.

How to structure the answer

• Situation, two or three sentences: the team, the goal, what was at stake and when.
• Task, one sentence: what you specifically were responsible for, or the problem in front of you.
• Action, most of your time: what you decided and did, why, and how you handled what got in the way. Say "I".
• Result: what changed, with a number where there is one (time saved, errors cut, users helped), and the outcome for the people affected.
• Learning, one or two sentences: what you took from it and what you now do differently.

Aim for about two minutes. Prepare six to eight stories from your own work that you can bend toward different questions: a hard technical problem, a disagreement, a failure, something you owned beyond your role, a time you influenced others, a time things were unclear.

An example answer

Question: "Tell me about a time you improved something that mattered to users." The details below are invented to show the shape.

Situation  Our checkout page took about six seconds to load on phones,
           and the team's goal for the quarter was fewer abandoned carts.
Task       I owned the page's performance work, with two weeks before
           the holiday code freeze.
Action     I profiled the page and found three API calls blocking the
           first render. I made two of them run in parallel, moved the
           third after the page appeared, and added a performance budget
           to our CI so the page could not quietly get slower again.
Result     Load time fell to about two seconds, and abandoned checkouts
           fell 9% over the following month.
Learning   I now set a performance budget on a page before I change it,
           instead of fixing speed after users complain.

Common mistakes

• Spending most of the answer on the situation and running out of time before the action.
• Saying "we" throughout, so the interviewer cannot tell what you did.
• Ending without a result, or with one nobody could check.
• Answers of five minutes or more, or of thirty seconds.
• Telling a prepared story that does not answer the question asked.

How to say it out loud

Signpost lightly ("The situation was...", "What I did was...") so the listener always knows where you are. Pause for a moment before you start; choosing the right story is worth two seconds of silence. Then listen to the follow-up: it usually asks for the part you rushed.

In Devana's behavioral rubric, 60% of the score is the content of your answers, and action and ownership is its largest single metric; the other 40% is delivery: clarity, listening, composure, pacing and structure. The fastest way to raise both is to spend more of each answer on what you did, in the first person.