STAR, without sounding like STAR

Situation, Task, Action, Result. Everyone knows it, most people use it, and it still produces bad answers — because STAR is a checklist of parts, and the thing that goes wrong is never a missing part. It's the proportions.

The setup is the easiest thing to talk about. You know it cold, it requires no self-assessment, and it feels like necessary context. So it expands, and it expands into the minutes that were supposed to hold what you did and what changed.

Where the four minutes go

A behavioural answer is a fixed-length container. STAR tells you which parts go in it and nothing about how big each part gets, and the division is where answers are lost.

Take a four-minute answer at an unhurried 140 words a minute — 560 words in total. Two ways to spend them. The first column is the setup-heavy answer that coaches keep describing; the second is what the rubrics' emphasis on action and outcome implies. Neither is survey data, because nobody publishes measured airtime splits; they are two shapes to compare.

The rubric-shaped split gives Result 2.5× the airtime. In absolute terms that is a difference of 84 words of Result — and 84 words is not a stylistic preference, it is the entire space in which a number, its source, and a consequence can exist. Under the setup-heavy split, the Result gets 56 words, which is enough for one vague sentence: "so it went really well and the team was happy."

The other number worth staring at is the you-and-outcome share: 40% versus 70%. In the setup-heavy answer, three out of every five words are about a situation that the interviewer cannot grade, because the situation is not the thing being scored. You are.

The three sentences that carry the score

Rather than rebalancing by feel, fix three specific sentences. Everything else can be improvised.

One sentence of Situation, ending in the problem. Not the history, not the reorg — the state of the world plus what was wrong with it. "We were migrating payments off a service that had a four-hour nightly window, and the migration had already slipped twice." That is 22 words and the interviewer now knows enough.

The Task sentence has to contain the word "I". This is the sentence where candidates go plural and lose the whole answer. "We decided to re-architect" is unscorable: the rubric asks what you did, and a story told entirely in "we" leaves the interviewer unable to write down a single thing you personally contributed. It is fine — good, even — to be one of several people. Say which one you were.

The Result sentence needs a number, a source, and a consequence. Three components, because each one is doing separate work:

If you genuinely don't have a number — plenty of real work isn't instrumented — say what you do have and say why. "We never measured it directly, but the on-call rotation stopped getting paged for that job, which had been three or four pages a week." That is honest, specific, and stronger than an invented percentage, which will not survive a follow-up.

Don't announce the framework

The last part of not-sounding-like-STAR is literal. Say the story, not the labels. "So, the situation was…" then "my task was…" telegraphs recitation and makes a four-minute answer feel like a form being read aloud.

You get the structural benefit — the interviewer being able to follow — from ordinary signposting instead:

"This was during the payments migration, about eighteen months ago. The short version is we had a four-hour nightly window that had already slipped twice, and I was the one who ended up owning the fix."

That does the work of S and T in one breath, is unmistakably a person talking, and lands the "I" without sounding like a claim of sole credit.

WatchProfessor Heather Austin · 06:50

A worked answer to the hardest of the standard questions. Jump to 03:38 for the delivery, then compare it against the budget above — the sample answer spends far more of its length on what changed than on the setup, which is the whole point of this lesson.

Tell Me About A Time You Failed — sample answer Play “Tell Me About A Time You Failed — sample answer” on YouTube

Surviving the follow-ups

The digging is the method, so plan for it. For every story in your inventory, have three layers ready:

1. The four-minute telling — what you'd say unprompted.
2. One level of technical or interpersonal detail — the actual mechanism. What the migration did, why the batch was four hours, what the other person's objection actually was. This is where assembled stories collapse.
3. The counterfactual — what you considered and rejected, and why. "We could have parallelised the existing job instead; I didn't because the row locks were the constraint, not the CPU." This layer is the one that most reliably distinguishes senior candidates, because it demonstrates the decision had alternatives.

If you cannot produce layer 2 or 3 for a story, it is a story you shouldn't lead with. Not because you're lying — because a thin story invites exactly the questions that make it look thin.

What's next

You can now shape a single answer. The next problem is which answers to have at all: five or six questions against eight signals, and a set of stories that probably clusters on three of them. That's a coverage problem, and it's solvable on paper before you're in the room.