Application process
The STAR Method for Civil Service Interviews (With Worked Examples)
Published · Last updated
The STAR method tells you the order of an answer and nothing about the weighting, and in a scored interview the weighting is nearly everything. Marks attach almost exclusively to the Action and the Result. Panels cannot score context; they score what you did, why, and what happened.
Most candidates deploy STAR in exactly the proportions that lose marks: a long scene-setting Situation, a Task restating the Situation, a rushed clump of Actions, and a Result of "it was delivered successfully and well received".
The weighting that actually matters
For a two-to-three-minute interview answer:
| Element | Share | What it has to contain |
|---|---|---|
| Situation + Task | ~20% | Two or three sentences. Who the work was for, what was at stake, and your ownership established in the first breath: "I owned", not "we were asked to". |
| Action | ~60% | Three to five distinct things you did, each carrying a because. This is where descriptors get ticked: the decision made, the option weighed, the person persuaded, the risk managed. |
| Result | ~20% | Quantified where possible, honest about scale, plus one grade-level reflection. |
Actions without reasoning are a to-do list. Reasoning is what turns them into judgement the panel can score.
The most common structural failure is not a missing letter. It is inversion: 60% situation, 20% action. Candidates feel the context is what makes the story impressive. Panels experience it as a minute of unscoreable throat-clearing while the descriptors sit unticked.
STAR vs B-STAR
You will see B-STAR promoted in coaching materials, adding a "Belief" sentence up front to frame what the example demonstrates.
Two things to know. It is not official: GOV.UK guidance recommends plain STAR, and no panel has a B on their scoresheet. It is occasionally useful anyway, as a discipline for you, forcing each example to target a specific behaviour, and as a signpost when a question is broad. But a belief statement scores nothing by itself, and a laboured one costs you seconds you need for Actions. If in doubt, use plain STAR, heavy on the A.
Some candidates also add L for learning (STARL). Same verdict: the reflection is worth one strong sentence inside your Result, not a section of its own.
Fail vs pass: the same events, re-weighted
A Delivering at Pace answer, first told with the classic inverted weighting.
FailVersion 1: structurally 'correct' STAR, scores around 3/7
"So the situation was that our team was responsible for migrating our public-facing guidance content onto a new website platform. This was part of a wider departmental digital programme that had been running for about a year, and our section covered around three hundred pages of content used by both the public and internal teams. The migration had a fixed cutover date because the old platform's contract was ending, and we were working with an external supplier who was handling the technical migration while we handled content checking. My task was to coordinate the content checking so everything was accurate and accessible on the new platform by the cutover date. About six weeks out, the supplier told us their part was running late, which compressed the time we'd have for checking. So what I did was: I re-planned our checking schedule, prioritised the most-used pages, got some extra help from another team, and we worked hard to get through it. As a result, we made the cutover date and the new site launched successfully with no major content issues, which the programme team were really pleased with."
Count the proportions: roughly 65% situation and task, 25% action, 10% result. The actions, when they finally arrive, come as an unreasoned list (re-planned, prioritised, got help, worked hard) with every scoring question unanswered. Prioritised how, against what? Whose help, obtained by what argument? What was dropped to make the compressed window work? The panel hears a well-described project and an invisible candidate. A 3, delivered in perfect STAR order.
PassVersion 2: the same six weeks, around 6/7
"I coordinated content checking for a 300-page migration to a new platform, with a fixed cutover — the old platform's contract was ending. Six weeks out, the supplier's technical work slipped, cutting our checking window roughly in half. Checking everything to full depth was no longer possible, so I made that explicit rather than letting quality quietly absorb the squeeze. I pulled twelve months of page analytics and split the content three ways: the fifty pages carrying over 80% of traffic got full checking including accessibility testing; the long middle got a lighter factual-accuracy pass; and about forty near-orphaned pages I recommended we not migrate at all — retiring them was faster than checking them, and the data showed they'd stopped earning their maintenance years ago. That recommendation needed sign-off, so I took it to the content owner with the analytics attached and one risk named honestly: a retired page might turn out to matter to someone, so we'd redirect rather than delete, reversible for six months. For the middle tier I borrowed two colleagues from a neighbouring team — their team lead agreed because I asked for a defined ten days with a page-count, not vague 'support'. We hit the cutover with the top tier fully checked; two retired pages were queried in the following month, both restored in minutes via the redirects. The retire-rather-than-migrate call cut the checking workload by about 15%, and the traffic-tiered approach became the template for the two directorates that migrated after us. What I took from it: when time is cut in half, the honest move is deciding out loud what gets less — not promising everything and letting standards slip in private."
Same six weeks, same supplier slip, same successful cutover. The proportions are now roughly 15/65/20, and every Action carries its reasoning: the analytics because checking depth had to be rationed, the retirement recommendation because the data showed the pages were not earning maintenance, the named risk with a reversible mitigation, and the bounded ask that got the loan of two colleagues. The Result has numbers, an honest wrinkle handled inside the mitigation, and spread. Structurally both answers are STAR. Only one of them contains a candidate.
Using STAR across the application
- Interview answers: the 20/60/20 weighting above, two to three minutes, then let probing questions pull out more. Probes are the panel hunting for marks, so treat them as bonus Action time rather than scepticism. Full detail in how Civil Service interviews are scored.
- 250-word behaviour statements: the same weighting compresses to roughly 40 words of situation, 150 of reasoned actions, 60 of result. On the page, ownership and reasoning have to be explicit because a sifter cannot probe. Worked examples in the 250-word behaviour statement guide.
- Live interview questions: the behaviour-by-behaviour question patterns STAR is answering are mapped in the interview questions hub.
- Personal statements: don't. They assess experience against the advertised criteria, and STAR's one-example depth is the wrong tool. Breadth mapped to the criteria wins. This is one of the most common self-inflicted sift failures.
Common mistakes that cap your score
- The inverted pyramid. Half the answer on context. If your Situation runs past three sentences, you are spending scoring time on scenery.
- Actions without a because. "I prioritised the workload" ticks nothing. "I prioritised by traffic, because checking depth had to go where the users were" ticks a descriptor.
- "We" throughout. Panels score you. Every "we did" is evidence donated to your team. Say what you did inside the team effort.
- The vanishing Result. One sentence of "it went well". Quantify, name what changed, add one line of grade-level reflection.
- STAR as a straitjacket. Announcing "the Situation was... my Task was..." out loud wastes time and sounds rehearsed. Use the structure invisibly.
- One story stretched over every behaviour. STAR makes examples portable, which tempts over-reuse. Panels notice, and the fit weakens each time. Build a bank of six to eight.
Common questions
Situation, Task, Action, Result — a structure for evidencing behaviours. In Civil Service interviews the weight belongs on Action and Result: roughly 20% situation and task, 60% action, 20% result.
A variant adding 'Belief' or 'Behaviour' at the front — a sentence framing the behaviour being evidenced. It's optional coaching shorthand, not an official Civil Service requirement. Panels score descriptors, not structure.
No. GOV.UK recommends it as a useful structure, and panels are used to it, but you're scored against behaviour descriptors, not adherence to an acronym. STAR is scaffolding, not the building.
Two to three minutes for a main answer. Long enough to land several scoreable actions with reasoning, short enough to leave room for probing questions — which are extra scoring opportunities.
No. Personal statements assess experience against the advertised criteria — breadth matters and STAR's single-example depth wastes the word count. Use STAR for behaviour statements and interview answers only.
Practise before the real thing
Reading how panels score is step one. The candidates who pass are the ones who practised saying their answers out loud — and got scored feedback against the official Civil Service Behaviours framework before interview day.
Start Practising FreeStart a practice session3 free practice questions · No card needed · Cancel anytime