The STAR method

1. Situation

Set the scene in two or three sentences. The interviewer needs enough context to judge how hard the problem was, and no more than that.

Include the size of the thing. "A team of five" and "about 40,000 orders a day" tell the interviewer more than "a large project" does. Leave out anything confidential, and anything about a colleague's private life.

A good Situation runs about 15 to 20 seconds.

2. Task

Say what you were responsible for. Candidates skip this part more than any other, and it is the part that makes the rest of the story mean something.

Compare two versions of the same sentence. The first: "our team had to fix the checkout failures." The second: "I owned the payment service, and the failures were in my code." Only the second tells the interviewer what was on you.

If you were not responsible for the outcome, say what you were responsible for instead. Being the person who noticed a problem is a real contribution. Claiming more than you did is the fastest way to fail a follow-up question.

A good Task runs about 15 seconds.

3. Action

This is the longest part and it should be. Say what you did, in order, and why you chose it.

Two rules make this section work.

Use "I", not "we". The interviewer is scoring you, and cannot score a group. Describe the team's work where it matters, then be exact about your own part.

Limit yourself to about three actions. Beyond that the interviewer loses track. Pick the three that mattered and drop the rest. You can always add detail if they ask.

Include one decision you made and the reason behind it. That is what separates someone who did the work from someone who was present while the work happened.

A good Action runs about 45 to 60 seconds.

4. Result

Say what changed, with a number.

This is where most answers fail. A candidate spends ninety seconds on a detailed story, then finishes with "and it went really well." The interviewer has nothing to write down.

Compare these two endings:

• "The team was much happier and the process improved a lot."
• "Release went from about a day to about 40 minutes, and we moved from weekly releases to daily ones."

The second is not a better story. It is the same story with the result stated. Numbers you can produce from memory are enough: how long something took before and after, how many people, how many incidents, what percentage.

Finish with one sentence on what you learned or would do differently. That sentence answers a follow-up question before it is asked.

A good Result runs about 20 to 30 seconds.

The most common failure

The problem is almost never that a candidate has not heard of STAR. It is that they abandon it two sentences into a real answer.

What happens is this. The Situation runs long, because context feels safe to talk about. By the time they reach the Action they can sense the answer has gone on too long, so they rush it. Then the Result becomes one vague sentence.

The fix is to time yourself once, out loud, on a single story. Most people are surprised how much of their answer turns out to be setup.

A worked example

Question: "Tell me about a time when you had to manage a project with tight deadlines and limited resources."

Situation: "I was one of two engineers on a checkout service handling about 40,000 orders a day. Users started reporting that checkout failed occasionally, and we could not reproduce it."

Task: "I owned that service. Every failed checkout was a lost order, and we were seeing about 300 a week. I had one week before our busiest sales period."

Action: "I started by adding logging around the transaction path rather than guessing, because two of us had already guessed for a day. That showed the failures clustered when two users checked out within a few milliseconds of each other. I reproduced it in a test, found a race condition in our database transactions, and added a lock so those transactions ran in order. I tested it in staging against replayed production traffic before shipping."

Result: "Failed checkouts went from about 300 a week to none over the following month. We went into the sales period without an incident. What I took from it is that a day of logging beats a day of theories. I now add instrumentation earlier than I used to."

That answer is about 90 seconds spoken. Every part is present, the Result carries a number, and the last sentence hands the interviewer a follow-up the candidate is ready for.