In this article
Before you start
Why interviewers ask this
An interviewer cannot watch you work, only hear how you describe it, and the STAR structure makes that description checkable. Google, for example, uses "the same interview questions" and grades responses "on an identical scale", and it documents "what a poor, borderline, solid, and outstanding answer would cover" (Google re:Work, updated March 2026). A structured interview had the highest mean operational validity (r = .42, a correlation between interview scores and later job-performance ratings) among the selection methods reviewed in a 2022 re-estimate (HumRRO summary of Sackett et al., 2022).
A weak answer is a general claim. A strong answer is one specific event with your own decisions inside it.
The STAR method from first principles
- Operational validity
- Operational validity is the correlation between interview scores and later job-performance ratings, corrected for measurement error; r = .42 means a moderately strong link, where 0 is none and 1 is perfect.
In plain words
A STAR answer works like a news report of one event. A report says where and when. It says what was at stake. It spends most of its length on what happened and ends with the outcome. It never opens with "I am a good communicator." The analogy breaks in one place: a reporter watches, but you acted. The Action must say "I" wherever you did the work.
The STAR method is a four-part order of answer. MIT's career office describes the parts as the Situation ("explain the situation so that your interviewer understands the context"), the Task ("the task that you took responsibility for completing or the goal of your efforts"), the Action ("the actions that you personally took") and the Result ("the positive outcomes or results generated by your actions"). It says most of the response should be Actions (MIT career-office guide, read October 2026).
Step 1: Situation, set the scene in two sentences
The Situation gives the interviewer just enough context to follow the story: what the team or product was, and what was going wrong or about to change. Its input is the event you chose; its output is two sentences. Name the setting in plain words, not internal project names the interviewer cannot know. Cut anything that does not change what you did later; a detail about the office layout or the project's history rarely does. MIT suggests about 20% of the answer for this part.1 What you now know: the interviewer can picture the scene without a long preamble, and you have not yet used the time the Action needs.
Step 2: Task, say what you owned
The Task states your responsibility or goal, which is different from the situation: the situation is what was happening, and the task is what you were answerable for. "The payment retries were creating duplicate charges" is a situation; "I was asked to stop duplicates without delaying legitimate retries" is a task. One sentence is usually enough. This step matters because the interviewer scores the Action against the Task: an action is only good relative to the goal it served. If the task was vague, say who set it and what done looked like. MIT suggests about 10% of the time.1 What you now know: there is a goal against which your action can be judged.
Step 3: Action, say what you did, in "I"
The Action is the longest part: the specific steps you took, in order, with the reason for each. Use ownership language wherever the work was yours, and say "we" only for what the group decided, because a follow-up probe will ask which part was yours. A follow-up probe is a question an interviewer asks after your answer to test whether a claim is real. Two or three concrete actions beat a list of six. Include one decision with a trade-off ("I chose a limit of three retries because a fourth rarely succeeded and raised the duplicate risk"). MIT suggests about 60% of the answer here.1
The weak version has no Task, no Action that belongs to anyone and no Result. Every interviewer follow-up it invites begins with "What did you do?" What you now know: the Action carries the score.
Step 4: Result, give a number or an observable outcome
The Result states what changed because of your Action, with a number or something observable. "Duplicate charges fell from about two a day to none for the next quarter" can be checked; "it worked well" cannot. If you do not have a number, say what you observed and when. Keep it to one or two sentences; MIT suggests about 10%.1 A result may be mixed, and saying so honestly is stronger than overselling, because an interviewer who probes will find the gap anyway and then doubt the rest of the story. What you now know: the story has an outcome the interviewer can weigh.
Step 5: Reflection, say what you changed afterwards
The reflection step is one more sentence that says what you do differently now because of the event. MIT folds learning into the Result ("You may also want to emphasize what you learned from the experience or your key takeaways"), so a separate closing sentence is this page's addition, not a fifth letter. It is the step the acronym leaves out and the sentence that most shows your level.
A result says what happened once; a reflection says it will happen differently next time. "Since then I add a limit and a written record to any retry logic before it ships" shows learning that carries to a new job. The sentence is short and specific, not a lesson about teamwork, and it should be something the interviewer could verify by asking how you work today. What you now know: the answer ends on judgement, not on a report.
Step-by-step description
- Set the scene in two sentences.
- Say what you owned.
- Say what you did, in I.
- Give a number or an outcome.
- Say what you changed afterwards.
- The time split follows MIT's percentages: 20% Situation, 10% Task, 60% Action, 10% Result.
- The reflection step is one more sentence that says what you do differently now because of the event.
How long should each part be?
The time split follows MIT's percentages: 20% Situation, 10% Task, 60% Action, 10% Result. For an illustrative two-minute answer of 120 seconds that is 24 seconds of Situation, 12 of Task, 72 of Action and 12 of Result; for 90 seconds it is 18, 9, 54 and 9; for 180 seconds it is 36, 18, 108 and 18.1
MIT says the percentages are a guide and that you should not worry about getting them exactly right. Add a few seconds for the reflection sentence. The totals are illustrative lengths, not measured limits. Time one rehearsal aloud with a timer and compare it with the bars below: a draft that spends more than a fifth of the time on the Situation is leaving the Action short.
Text description of the diagram
- For an illustrative two-minute answer of 120 seconds that is 24 seconds of Situation, 12 of Task, 72 of Action and 12 of Result.
- For 90 seconds it is 18, 9, 54 and 9; for 180 seconds it is 36, 18, 108 and 18.
Knowing it is half the job. Saying it is the other half.
Interview Clozo is an AI interview copilot for real interview rounds: it hears the question and puts a strong answer on your screen, and you say it in your own words. See the features or how it works.
Start Free Trial →The Staff-level answer
A Staff-level answer keeps the STAR order but chooses a story with real stakes, makes one decision visible, and ends on a change in how you work. Below, an illustrative composite answer to the question "Tell me about a time something you owned was failing in production."
Tell me about a time something you owned was failing in production.
Our payment service was charging some customers twice when a bank timed out, and each duplicate meant a refund and a support ticket. I owned the retry path, so the fix was mine. I reproduced the duplicates from the logs, then gave each charge one idempotency key that every retry reuses, so the payment provider treats a repeat as the same charge.
I limited retries to three, because the logs showed a fourth rarely succeeded. The team wanted to ship the key alone; I made the call to add the limit too, accepting that a few slow banks would fail sooner. Duplicate charges went from about two a day to none for the next quarter. I then wrote the review rule that payment calls need a key and a retry limit, and other payment teams adopted it.
For comparison, here is a mid-level answer to the same question, an illustrative composite.
The mid-level answer has a Situation, an Action and a Result, so it passes the four letters. It has no stated decision, no number and no change in practice, so an interviewer who wants to score judgement has nothing to score. The Staff-level answer is 137 words and the mid-level one is 40, so the gap is 97 words, which is why the extra sentences are worth rehearsing.
What makes the Staff-level answer different: it names the decision and its evidence, states the cost of the problem and of the trade-off, separates your call from the team's, quantifies the result, and ends on a rule that reached other teams. Say the structure to yourself as: scene, ownership, my actions and a decision, outcome, what I changed.
The rubric
| Dimension | Weak answer | Strong answer | Staff answer |
|---|---|---|---|
| Specificity | A general claim ("I usually...") | One dated, specific event | An event with stakes and a decision |
| Ownership | "We" throughout | "I" for your own actions | Separates your decision from the team's |
| Result | "It went well" | A number or an observable outcome | A number plus what it cost or risked |
| Reflection | None | A lesson | A changed practice you now apply |
| Follow-up readiness | Surprised by "What did you do?" | Can expand each step | Expects the probe and answered it already |
Three variants of this question
"How do you use STAR if you have little work experience?"
Choose the event from anywhere you acted: a course project, a volunteer role, a part-time job or a club. The structure does not need a job title, only a real situation, a task you owned and actions that were yours. MIT's own sample STAR answer comes from school: a high-school junior who helped classmates struggling in a math class (MIT career-office guide, read October 2026). Keep the scale honest: a small event told specifically beats a large one told vaguely.
"How do you use the STAR method for a question about failure?"
Use the same order, but make the Result the honest cost and the Reflection the change. A failure story with a clear reflection is often stronger than a success story without one, because it shows you can judge your own work. Choose a failure that was real but recoverable, state your share of it without blaming a colleague, and spend the Reflection on a concrete change such as a check you now run.
"Does STAR work if the question is not about the past?"
Only partly. Google distinguishes behavioral questions, which ask about a past situation, from hypothetical ones, which present "a job-related hypothetical situation" ("What would you do if...?") (Google re:Work, updated March 2026). For a hypothetical, give your approach and its trade-off, then offer a real example that shows you have done it.
What are common STAR method interview questions?
Most behavioral questions start "Tell me about a time when..." and test one theme. The table maps ten common prompts to the story theme each one needs; the prompts are illustrative, not a list from any one employer.
| Common prompt (illustrative) | Story theme it needs |
|---|---|
| Tell me about a time you failed. | A real, recoverable failure and what you changed |
| Tell me about a conflict with a colleague. | Disagreement handled without blame |
| Tell me about a time you led without authority. | Influence with no title |
| Tell me about a time you worked under a tight deadline. | Prioritising and cutting scope |
| Tell me about a time you made a mistake. | Owning an error early |
| Tell me about a time you disagreed with your manager. | Respectful challenge, then commitment |
| Tell me about a time you learned something quickly. | A learning method, not just a result |
| Tell me about a time you improved a process. | A measurable before and after |
| Tell me about a time you handled an unhappy customer. | Listening, then a fix |
| Tell me about a time you took initiative. | Spotting a gap nobody assigned you |
Three short composite outlines show the five notes for different fields. They are illustrative examples, not real people.
Illustrative example, non-engineering (retail shift lead): Situation: Saturday queues ran past 20 minutes at a shop I supervised. Task: I was asked to shorten them without adding staff. Action: I moved one person from stock to a second till at peak hours and put the busiest returns in a separate line. Result: waits stayed under ten minutes on the next three Saturdays, from my own count. Reflection: I now check the busiest hour before changing a rota.
Illustrative example, engineering (slow page): Situation: our signup page took six seconds to load. Task: I owned getting it under two. Action: I measured first, found one oversized image set, resized it and deferred a third-party script. Result: the page loaded in under two seconds on the same test connection. Reflection: I now measure before I optimise.
Illustrative example, student (group project): Situation: our course group had one week left and no finished report. Task: I volunteered to coordinate. Action: I split the report into four sections with names and dates and held two ten-minute check-ins. Result: we submitted a day early. Reflection: I now set owners and dates in the first meeting.
Common mistakes
- Spending the time on the Situation. (wrong) "Let me explain the history of the project first." It uses the minutes the Action needs. (right) Give two sentences of scene, then move to what you owned. As a rule of thumb, the first "I" should arrive by the end of the Situation, about a fifth of the way into the answer.
- Saying "we" for everything. (wrong) "We decided to add a limit." The interviewer cannot tell what you did. (right) "I proposed the limit and the team agreed." If the work really was shared, say which slice was yours and which decision was the group's.
- Ending on the Result. (wrong) "And it worked." A report with no learning. (right) Add one sentence: "Since then I check for X before shipping."
- Telling a story with no number or outcome. (wrong) "It went pretty well." (right) "Duplicates fell from about two a day to none for the next quarter."
- Reciting a memorised script. (wrong) A polished speech sounds rehearsed and breaks on the first probe. (right) Memorise the outline, not the sentences. Write each story as five short notes, one per step, and speak from the notes so a probe does not derail you.
Follow-up questions
How do you use the STAR method?
The STAR format is the same thing as the STAR method: Situation, Task, Action, Result. Pick a true story that fits the question, say the Situation in two sentences, state your Task, spend most of the time on your own Actions, give the Result with a number, and close with what you changed. Practise aloud against a timer; the tell me about yourself page shows the same time-budget thinking for an opener.
Why is the STAR method effective?
It matches how structured interviews are scored: the interviewer wants a specific past event and checks it against a written scale, and the four parts map to what the scale looks for. Structured interviews had the highest mean operational validity in the 2022 re-estimate cited above, so the method is practice for the format with the highest mean validity in that re-estimate. It also helps the interviewer: a candidate who answers in order gives a scorer something to check against each line of the scale.
How many STAR stories should you prepare?
MIT recommends 3 to 5 stories (MIT career-office guide, read October 2026). Choose them so that each can answer several themes, such as conflict, failure or leading without authority. A practical test is to write the theme of ten common questions on paper and check that each one maps to at least one story; if two stories cover everything except failure, the missing story is the one to prepare next.
Can the same story answer different questions?
Yes, by changing which Action and Result you emphasise. A production-incident story can answer a question about failure, ownership or pressure; what changes is the part you stress and the Reflection you end on. The next pages in the path are questions to ask the interviewer, the wider interview loop and salary negotiation; managers can see the same structure at scale in leadership interviews and engineering management interviews.
What if you have no number for the Result?
Say what you observed and when: "the complaints stopped by the next review", "the release went out without a rollback". An observable outcome that the interviewer could ask about is acceptable; an invented figure is not, because a probe will test it. If you can, name how you would measure it now.
What if your best example is not from work?
Use it, as long as the situation was real and the actions were yours (see the variant above on little work experience). Pick the example where your decision is easiest to describe, and keep the same five notes.
- STAR is Situation, Task, Action, Result: one real event, in that order.
- Spend about 60% of the answer on your own Actions, in "I".
- Give the Result a number or an observable outcome.
- Add one reflection sentence: what you changed afterwards.
- Prepare 3 to 5 stories as outlines, not scripts.
Sources
- Using the STAR method for your next behavioral interview — MIT Career Advising and Professional Development (date = read date), 24 Oct 2026
- A guide to structured interviewing for better hiring practices — Google re:Work (updated March 2026), 1 Mar 2026
- Is cognitive ability the best predictor of job performance? New research says it's time to think again — HumRRO (summary of Sackett et al., 2022; read 2026-10-24), 27 Sep 2022
