Very few candidates know how their Civil Service application is actually marked, which is strange, because the scoring approach is not secret and knowing it changes how you write. People assume the panel forms a general impression. In reality they are scoring each Behaviour separately against a scale, usually 0 to 7, and then adding up. That has practical consequences: a brilliant answer to one Behaviour cannot rescue an empty answer to another, and a consistently solid application beats a spiky one.
This guide covers the scale, where the bar tends to sit, and the specific differences between an answer that scores in the middle and one that scores well.
The scale
Most Civil Service recruitment uses a seven point scale, applied to each Behaviour or Strength being assessed:
| Score | Label | What the panel saw | |---|---|---| | 7 | Outstanding | Compelling evidence at or above the required level throughout | | 6 | Excellent | Strong evidence across the whole descriptor, few gaps | | 5 | Good | Solid evidence, meets the level clearly | | 4 | Acceptable | Meets the level, some parts thin | | 3 | Minimally acceptable | Partial evidence, gaps against the descriptor | | 2 | Poor | Little relevant evidence | | 1 | Very poor | Evidence contradicts the requirement | | 0 | No evidence | Nothing relevant offered |
A few important caveats. Departments and agencies do vary, and some use a shorter scale or slightly different labels, so always read the advert and any candidate pack. Some campaigns also set a minimum score on individual elements, meaning you can fail on one Behaviour even with a strong total. If the pack states a minimum, treat it as a hard gate rather than an average.
Where the bar usually sits
In practice, 4 is commonly the minimum acceptable score, and on competitive campaigns the realistic threshold to progress is higher. For popular schemes and oversubscribed adverts, candidates scoring straight 4s frequently do not make the sift, because the cut is drawn wherever the volume requires it.
This is the most useful thing to internalise: scoring "acceptable" everywhere is often not enough. The aim is not to clear the bar on each Behaviour, it is to give the panel a reason to score you 5 or 6. Understanding what creates that difference is most of the work.
What separates a 4 from a 6
The gap is rarely the quality of the underlying experience. It is almost always in what the candidate chose to include. Four differences account for most of it.
Specificity. A 4 says "I improved the process and it saved time". A 6 says what the process was, what specifically was wrong with it, what change was made, and how much time it saved. Panels can only mark what is on the page, and vague claims are marked as unevidenced rather than assumed generous.
Evidence against the whole descriptor. Each Behaviour descriptor has several bullets. An answer covering one bullet well and ignoring the others reads as partial and lands at 3 or 4. An answer that touches each element of the descriptor reads as complete. This is the single most mechanical way to raise a score: read the descriptor, and check your answer speaks to every part of it.
Your reasoning, not just your actions. A 4 describes what happened. A 6 explains why you chose that course, what you weighed, and what you decided against. Making Effective Decisions is the obvious case, but reasoning lifts every Behaviour, because the framework is about how you work.
A result with a number and a consequence. Not every outcome is quantifiable, but most have something measurable: a percentage, a time saving, a volume, an error rate, a satisfaction score. Where you genuinely cannot quantify, give a concrete consequence instead: the decision that was taken, the complaint that was avoided, the practice that was adopted elsewhere.
A worked comparison
Same underlying experience, written two ways, for Communicating and Influencing at Level 3.
Likely a 3:
"I had to explain a new process to my team. Some of them were resistant to it. I communicated the benefits clearly and answered their questions, and eventually everyone got on board with the new way of working, which made the transition much smoother."
Nothing here is untrue, but the panel has nothing to mark. There is no specific audience, no named objection, no method, and no evidence of influence as opposed to announcement.
Likely a 5 or 6:
"I had to bring a team of nine caseworkers onto a new triage process, and two of the most experienced were openly against it, which mattered because the newer staff took their lead from them.
Rather than present the benefits to the whole team, which I judged would harden their position in public, I spoke to both of them individually first and asked what specifically worried them. The concern was not the process itself, it was that the new triage codes did not cover a category of case they handled often, so they expected to be forced into recording things inaccurately.
That was a legitimate gap I had not spotted. I took it to the process owner with their examples, and we agreed an additional code before rollout. I then asked one of them to walk the team through the triage change at our next meeting rather than doing it myself.
Adoption was complete within two weeks with no escalations, and the additional code was used on around fifteen per cent of cases in the first month, which confirmed the gap was real. What I took from it was that resistance from experienced staff is usually information rather than obstruction, and treating it that way was faster than persuading."
The second version scores because it names a specific audience and objection, shows a deliberate choice of method with a reason, demonstrates actual influence in both directions, and closes with a quantified result plus a reflection. It is longer, but not padded: every sentence adds something markable.
Practical implications for how you write
Write to the descriptor, not to the word count. Adverts often give a limit of 250 or 500 words per Behaviour. Use it, but spend it on action and reasoning. A long scene setting paragraph is the most common way candidates waste half their allowance.
Weight your effort evenly. Because each Behaviour is scored separately, your weakest answer drags the total more than your strongest lifts it. If you have four Behaviours to evidence and one feels thin, the highest return on your time is fixing that one rather than polishing the best.
Do not assume the reader knows your context. Sifters may be from a different department entirely. Acronyms and internal team names cost you marks by making the evidence harder to follow.
Expect the interview to probe the same evidence. At interview the panel often works from what you wrote, asking follow up questions on the same examples. Anything you exaggerated on the form becomes uncomfortable in the room, which is a practical reason to keep your written evidence accurate.
What this means for the interview stage
At interview the scoring works the same way, with each Behaviour or Strength marked separately. The difference is that you are speaking, so structure carries more weight: a panel taking notes needs to be able to tell when you have moved from situation to action to result. The STAR structure exists for exactly this, and it is worth using explicitly rather than subtly.
Strengths are scored on the same scale but assessed on a completely different basis, since they are about what you naturally enjoy and do well rather than evidence of past achievement. That distinction is covered in the Civil Service Strengths guide, and it catches out candidates who try to answer Strengths questions with prepared Behaviour examples.
For the framework as a whole, including how Sift and Interview fit together, see the Civil Service interview guide and the full breakdown of the nine Behaviours.
The most reliable way to find out whether your answers would score a 4 or a 6 is to deliver them out loud and hear where the evidence thins out. You can practise Civil Service style questions and get structured feedback on exactly that before you are in front of a panel.