Four letters, and what each is doing in the sentence
STAR stands for Situation, Task, Action, Result. The Situation is the context — where you were, what
the system or team was, why anything needed to happen. The Task is your specific responsibility
inside it, which is the letter most people skip and the one that tells the interviewer whether you
were the protagonist or a bystander. The Action is what you personally did, in sequence, including
what you decided against. The Result is what was different afterwards, stated as a change from a
baseline.
The useful way to understand it is that STAR was never designed as a way for candidates to speak. It
is a shape for interviewers to take notes in, so an hour of conversation reduces to something a panel
can compare across five candidates. Deliver an answer in that order and you are filling in their form
for them; deliver it in some other order and they have to reconstruct it while you are still talking,
and anything they cannot place gets dropped.
Why the shape exists at all
An interviewer has to write down a rating and defend it in a debrief, and "strong collaborator" is
indefensible on its own. What survives is a quoted specific: he pushed back on the schema change,
proposed running both versions for a fortnight, and drift went from about fifty a day to under
three. The quotable unit is an action plus a result. STAR is the minimum scaffolding that
guarantees both are present, and behavioural interviewing adopted it because unstructured
conversation reliably produced answers with neither.
It is also why the interviewer will interrupt you. Someone cutting in with "sorry, what did you
do there" is not being rude; they are missing a field and need it before the hour runs out. Treat
every interruption as a hint about which letter is thin.
Rough proportions, and why they are counter-intuitive
Over about two minutes, a workable division is fifteen seconds of Situation, ten of Task, sixty to
seventy of Action, and twenty to thirty of Result. Almost nobody does this naturally: the Situation
is the part you understand best and feel obliged to explain, so it expands, and by the time you
reach what you did the interviewer has stopped writing. The discipline is to give only the context
that makes your action intelligible and trust that anything else will be asked for. "We had a
nightly reconciliation job that kept failing and nobody owned it" is enough Situation. It does not
need the acquisition that produced the job or a tour of the schema.
Three ways it comes apart
All Situation, no Action. The commonest failure by a wide margin: ninety seconds of context,
then "so we fixed it and things got better". Diagnostically, count the sentences whose subject is
you doing something specific. Fewer than four and the answer has no Action section, regardless of
how long it ran. The repair is mechanical — for each story, write in advance the four to six
sentences beginning with a verb you performed: I traced, I proposed, I rejected, I
sequenced, I walked two on-call engineers through.
"We" where "I" belongs. Candidates drift into the plural for two honourable reasons: the work
genuinely was collective, and claiming individual credit feels graceless. But the rubric has a line
for your contribution and nobody can fill it from a plural pronoun, so an answer entirely in "we"
scores as no evidence rather than as modesty. The resolution is not to lie about the team, it is to
name the team and then narrow to your slice, which reads as more senior rather than less: "There
were four of us. Priya built the receiver; I owned the migration sequence and made the call to run
both paths in parallel." Having said that once, the singular is now accurate and you can use it
freely.
No Result, or a Result with no baseline. "It went much better after that" is not a result.
Neither is "we improved performance by 60%", because nothing has been stated to have been 60%
better than. A defensible result carries the before value, the after value, and how it was
measured. Where you genuinely never instrumented it, do not manufacture a percentage — switch to
scope, which is always measurable: four services migrated, three teams moved off the old endpoint,
six weeks of manual work removed. An honest scope figure outscores an invented improvement figure,
because the invented one dies on the follow-up "what was it before?".
The failure that is harder to see
An answer can hit all four letters and still fail, if it was memorised as a paragraph. Recited
answers are audible — the cadence flattens, the clauses are too well-formed — and they break in a
particular way: asked something from the middle, the candidate restarts from the Situation or loses
their place. Store facts rather than sentences. The fork you took, the option you rejected, who did
what, the before number, what changed afterwards. Those five regenerate the wording differently
every time, which is what you want, because the same story will have to serve several different
questions.
A worked pattern to adapt
The point of showing this is the shape and the ratios, not the words. Load your own facts in.
Situation. "Our nightly job that reconciled subscription state against the payment provider had been failing about twice a week for months, and no team owned it." Task. "I picked it up on the understanding that I was making it reliable, which is what I had been asked for." Action. "I pulled three months of failure logs and concluded the batch shape was the problem rather than the code, because the failures clustered where the provider rate-limited us.
So I proposed webhook-driven updates with a daily sweep kept as a safety net. That was contestable — it turned a two-week fix into roughly two months — so I wrote it up with the failure data and took it to my lead rather than just doing it.
Three of us built it; I owned the design and the cutover, and I insisted on running old and new side by side for three weeks comparing outputs, which caught a currency-rounding difference that would otherwise have reached customers." Result. "Drift, counted as accounts disagreeing with the provider at the daily sweep, went from between forty and sixty a day to typically under three over the two months after cutover.
I also turned the comparison harness into a permanent daily check, because otherwise we would only have learned about the next regression from a customer."
Roughly forty words of setup against a hundred and fifty of action and outcome, and every follow-up
an interviewer might ask has a hook already in it: the rejected alternative, the escalation, the
named colleague, the harness.
Where STAR stops helping
It is a floor, not a ceiling. It gets a complete answer out of you; it does not demonstrate
judgement, which is what senior loops are grading. Judgement lives in the material you choose to put
inside the Action — the tradeoff you named, the thing you deliberately did not do, the risk you
accepted knowingly. A perfectly shaped answer about work that required no decisions still reads as
junior.
STAR is the interviewer's note-taking form; your job is to fill in the Action and Result
fields with things only a participant could know, and to keep the Situation short enough that
you get there while they are still writing.
© 2026 Preptima. Originally published at preptima.com (opens in a new tab).