Changing and Improving: Civil Service Examples That Actually Score

Most Changing and Improving answers fail at the same sentence, and it's usually delivered with pride: "I suggested this to my manager, who agreed and took it forward." The candidate hears initiative. The panel hears the exact moment ownership left the story. From HEO upward this is the single most common cap on the score — the suggestion trap — and it's brutal precisely because suggesting improvements is genuinely the right thing to do at work. The interview doesn't score what was right; it scores what the descriptors describe, and from HEO the descriptors describe someone who works with others to identify and implement improvement, not someone who feeds ideas upward.

What panels are actually scoring

Strip the definition back to what earns marks and four things remain:

A problem you noticed, quantified. Not "the process was inefficient" — how inefficient, costing what, to whom. The gap between "it was clunky" and "it was costing the team roughly a day a week" is often the gap between a 3 and a 5, because the second version proves you reviewed ways of working rather than just experienced them.

Ownership through implementation. The idea is the cheapest part of the story. Marks sit in what happened after: who you tested it with, what broke, what you adjusted, how it stuck. If your example ends where implementation begins, so does your score.

Managed risk. At HEO/SEO the descriptors say "take managed risks by fully considering the varied impacts" — meaning the panel wants to hear the downside you weighed, and what you did to contain it. A pilot, a fallback, a deliberate limitation of scope. Change with no considered risk reads as either luck or exaggeration.

People who didn't want the change. Descriptors from EO up talk about helping others understand change and supporting them to adapt. A resistor you brought along is worth more than a room that clapped.

The questions panels actually ask

• "Tell me about a time you identified and implemented an improvement to a way of working."
• "Describe a change you made that others were resistant to."
• "Give an example of when you used technology to make a process more efficient."
• "Tell me about a time you had to adapt to a change you didn't agree with." (Note the reversal — your evidence here is constructive challenge plus professional implementation.)
• "Describe an improvement you made that didn't go to plan."

Probing follow-ups: "Whose approval did you need?", "What could have gone wrong?", "How did you know it worked?" — all three aimed at the ownership, risk and evidence marks above.

Fail vs pass: the same improvement at 3 and at 6

A realistic answer at HEO level. The events don't change between versions; the telling does.

Read it as a panel member. A real problem, genuinely fixed, and the candidate did most of the labour. But trace the decisions: the go-ahead was the manager's, the review was the manager's, the rollout was the manager's. The candidate's own contribution is executed within someone else's ownership — "asked me to help" is doing quiet, fatal work in that second sentence. There's no measurement on either side (how long did induction take before? after?), no risk considered, nobody resistant, and the closing evidence is praise rather than outcome. Sincere, useful, and capped at a 3 from HEO up.

What changed, ingredient by ingredient:

1. "I decided to fix it rather than flag it." One sentence relocates the ownership. The manager still appears — at the end, presented with a finished pack and evidence, which is where a manager belongs in an HEO answer.
2. A baseline. Six weeks to independent working, timed. Now "under four weeks" at the end means something. Improvement claims without a "before" number are just adjectives.
3. Risk, identified and designed against. The staleness point is the managed-risk descriptor made concrete — the candidate names why the last version failed and builds the countermeasure in.
4. The resistor, converted rather than overridden. The sceptical colleague's objection is specific, taken seriously, and absorbed into the design. That's "encourage ideas for change from a wide range of sources" scored in a single move.
5. A trial before adoption. Two starters, then adopt — a managed rollout, not a launch and hope.
6. Spread. Two other teams copying the format is the difference between "my improvement worked" and "my improvement created value beyond me."

How Changing and Improving changes by grade

• EO: reviewing your own and the team's ways of working, suggesting improvements and helping others adapt is genuinely enough — the suggestion trap doesn't fully bite yet. What separates a 5 from a 3 at EO is specificity and follow-through. The EO guide calibrates this in detail.
• HEO: the trap's prime territory. "I suggested to my manager" caps the answer; the panel wants identification and implementation, working with others, with the manager repositioned to sign-off. Full calibration in the HEO guide.
• SEO: your improvement should touch more than your own team's convenience — simplified processes, fewer resources, managed risk weighed across different end users, and an environment where others feel safe to challenge. The resistor matters more here, not less. See the SEO guide.
• G7: the descriptors shift to creating the conditions for improvement — space for others' ideas, constructive challenge to senior management on change proposals, early detection when change is going wrong. An improvement you personally executed, however good, is SEO evidence at G7. The G7 guide covers the shift.

Common mistakes that cap your score

1. The suggestion trap. Covered above. If your manager approved the idea before you did anything, restructure the telling so your ownership is visible — or pick a different example.
2. No baseline. "It's much better now" scores nothing. Measure before, measure after, even roughly.
3. The idea as the achievement. Spending 80% of the answer on how you spotted the problem and 20% on fixing it inverts where the marks are.
4. Zero risk. If nothing could have gone wrong, the panel can't score managed risk. Name the downside you weighed.
5. Unanimous applause. Everyone loving the change immediately is a missing-evidence problem. Find the sceptic; they're worth a mark.
6. Technology name-dropping. "We started using a new tool" isn't the technology descriptor. The mark is efficiency you can quantify because of a tool you chose deliberately.

---

The pattern in this article — quantify, own, manage the risk, convert the resistor — applies across every improvement story you have. If you want to pressure-test yours before the panel does, the AI interview coach scores your actual answers against the grade-level descriptors and shows you exactly where the marks are leaking. Three practice sessions are free, and membership adds the full question bank and every grade guide.