How to Answer Civil Service Behaviour Questions

A Civil Service behaviour answer scores on three things: what you owned, what you decided and why, and what changed as a result. Everything else in the story is context, and context is unscoreable.

That is why good careers produce bad answers. The instinct is to explain the situation properly so the panel understands how difficult it was. The panel cannot award marks for difficulty they understand; they can only award marks for descriptors you evidence.

The shape that fits the scoring

Two to three minutes spoken, or 250 words written unless the advert specifies otherwise. Within that, roughly 20% context, 60% your actions and the reasoning behind them, 20% result.

The 60% is the part candidates under-build. An action without a reason is a task; an action with a reason is judgement, and judgement is what the descriptors from EO upwards actually ask for. "I rescheduled the reviews" is a task. "I rescheduled the reviews rather than the audit because the audit had a statutory date and the reviews did not" is the same sentence carrying a mark.

The 20% context is the part candidates over-build. Two sentences: what the work was, and what made this instance hard. The STAR weighting covers why the conventional proportions get inverted and what that costs.

Start with ownership, not with scene-setting

The first sentence should establish that this is your story.

Compare "Our team was responsible for clearing a backlog of applications" with "I was responsible for a caseload of around 200 applications when the backlog doubled." The second has told the panel who is being scored before the story starts, and it costs no extra words.

This matters more than it seems because of how panels work. They score only stated evidence and are not permitted to infer. A "we" covering something you personally did is evidence handed to a team that does not exist on the scoresheet.

I and we, used deliberately

The rule is simple and the application is not, because over-correcting sounds false.

Use "we" to establish context: who the team was, what it was collectively for. Use "I" for every decision, action and judgement. If a colleague genuinely did something, say so; a candidate who claims all of a team effort reads as unreliable, and the descriptors for Working Together and Leadership need other people in them.

The test to apply to a draft: highlight every verb whose subject is "we", and ask of each whether you personally did that thing. Where you did, change it. Where you did not, leave it and add what you did instead.

Put the friction back in

Most prepared examples have had the difficulty edited out of them, because the instinct is to present competence.

A story where the plan worked, nobody objected and nothing was traded off cannot tick the mid-grade descriptors, which is why smooth answers cap around a 3 or 4 however impressive the outcome. What panels are scoring is what you did when it was not obvious what to do.

So the examples to choose are the ones with a constraint, a disagreement, a resource you did not get, or a decision that turned out to be wrong. Especially the last: an answer that includes what you would do differently evidences self-awareness that scores rather than weakness that costs.

Land a result you can measure

"It went well and was well received" is a claim. "We cleared the backlog eleven days before the deadline and the error rate held at under 2%" is evidence.

The number does not need to be impressive. It needs to exist, because a measured result is the difference between the panel believing you and the panel being able to tick something. Where there is genuinely no metric, a specific qualitative outcome works: the process was adopted by the other two teams, the complaint was withdrawn, the policy changed.

Written answers differ in one way

The construction is identical; the difference is that a sifter cannot probe.

At interview, anything you leave implicit can be pulled out by a follow-up question. On the page nothing can, so every element has to be explicit in the first telling: ownership, reasoning, result. That is why 250-word statements reward density in a way spoken answers do not, and why a good spoken answer transcribed usually scores lower than it sounded. Worked 250-word statements show the compression in practice.

Treat probes as extra marks

When a panel asks "what was your specific role in that?" or "why was that your decision to make?", they are not sceptical. They are looking for a descriptor your answer did not tick, and giving you a second chance at it.

Answer with new specificity rather than by restating. If the probe repeats, the panel still has not heard what it needs, which usually means they want the reasoning behind a decision you described as though it were obvious.

The self-check before an interview

Run each prepared example against five questions. If any answer is no, that example is currently scoring below what it could.

1. Does the first sentence say what I was responsible for?
2. Is there a decision in this that was mine, with the reason attached?
3. Did something go wrong, get contested, or have to be traded off?
4. Is there a number or a specific outcome at the end?
5. Could someone one grade below me have given this same answer?

The fifth is the one that catches strong candidates, and it is the hardest to see from the inside. If the answer is yes, the example is pitched low, which is the most common single cause of a score of 3 rather than a 6. Why answers score 3 works through the diagnosis, and how to prepare for the interview overall puts this step into a two-week plan.