UK public sector applications
Most strong private-sector candidates fail the public sector sift for one reason: the panel can only score evidence that is written down against each criterion, in a form it can match. A brilliant CV that says "drove $3M ARR" scores poorly against "delivering business value through financial management" unless someone does the translation. This skill is that translation, done honestly.
Non-negotiables
- Never invent or inflate. Every claim in an output must trace to something in the CV, the user's own words, or an answer they gave. Panels probe claims at interview and SC/DV vetting checks history; padding gets found. Where evidence is thin, say so and ask, rather than smoothing it over.
- "I", not "we". The panel scores the candidate's own actions. Convert team achievements into what the candidate personally decided, led or did.
- Mirror the criteria. Use the advert's own wording as headings or opening lines so the sifter can tick each criterion without hunting.
- Numbers carry scale. People led, budget held, users served, uptime, cost saved, time cut. Scale is how panels judge grade fit.
- British English, plain language. No private-sector jargon left untranslated (see
references/translation-guide.md).
Workflow
1. Gather inputs (ask only for what is missing)
Needed: the advert or candidate pack; the CV; the grade; what the sift assesses (CV, personal statement, behaviour statements, experience, technical); word or page limits; the closing date. Useful but optional: anything the user knows about the team, and whether they want to apply under the Disability Confident Scheme or need reasonable adjustments (their decision; just make sure they know the option exists if the pack mentions it).
If a word limit is not stated anywhere the user can see, say so and ask them to check the online form; do not guess silently.
2. Decode the advert
Extract into a table: each essential criterion, each desirable criterion, the behaviours listed, the grade, and at which stage each is assessed (sift, test, SEE, interview). Note the "lead criterion" or "lead behaviour" if the advert says large application volumes will be sifted on it first — that one gets the strongest example.
Identify the framework:
- Civil Service and most departments/agencies → Success Profiles. Read
references/success-profiles.mdand use the level matching the grade (seereferences/grades.md). - NHS, police, local government, devolved bodies → read
references/other-frameworks.md.
3. Build the evidence bank from the CV
For each role, pull out candidate examples with: situation and scale, the candidate's personal action, the measurable result, and dates. Keep the user's numbers exactly as written. Note which facts are vague (e.g. "led a team" with no size) — those become questions.
4. Map evidence to criteria — show this before drafting
Produce an evidence map using references/evidence-map-template.md: for each criterion, the best one or two examples, a rating (Strong / Partial / Gap), and for anything below Strong, a specific question that could close the gap ("Did the DevOps team build self-service environments or golden paths for product engineers? If so, how many teams used them?").
Also give a short, honest grade-fit read: where the evidence clearly meets the grade descriptors, and where a panel is likely to probe (scale of line management, budget size, operational accountability). The user decides whether to apply; the job here is to make sure they decide with clear eyes and that the application addresses the weak spots head-on.
Wait for the user's answers before drafting when there are Gaps on essential criteria — their answers usually contain the best material.
5. Translate
Apply references/translation-guide.md: reframe revenue as value for money and service outcomes, customers as users/citizens, CEO reporting as senior leadership and governance, and so on. Add public-sector context only where the candidate genuinely engaged with it (e.g. HMRC Making Tax Digital work), never as decoration.
6. Draft
Choose the structure the sift needs:
- Personal / supporting statement against essential criteria (common at SCS and many G7/G6 roles): use
references/statement-template.md— short opening on motivation and fit to the organisation's mission, then one headed section per essential criterion, each with one main example and, where useful, a second supporting line from a different organisation. Panels for senior roles explicitly like evidence "from a range of roles in different organisations". - Behaviour statements (usually 250 words each, delegated grades): one STAR example per behaviour, pitched at the grade's level descriptors. Roughly: Situation and Task 20%, Action 60–70%, Result 10–20%. Action is what scores.
- CV for the sift: reverse chronological, two pages unless told otherwise, achievement-led bullets that echo the criteria, explicit reasons for any gaps in the last two years if the pack asks, no photo, nothing protected (age, marital status). Civil Service sifts are often name-blind — keep identifying detail out of statements unless the form says otherwise.
Split words across criteria in proportion to how much the panel will weigh them and how much evidence needs bridging, not equally by default.
7. Check before handing over
Run through references/quality-checklist.md. Report the word count per section and in total. List any claims the user should double-check (dates, figures) and any criterion still only Partial.
8. Next stages (offer, don't push)
After submission the same evidence bank feeds interview prep: blended interviews (behaviours + experience + technical), Staff Engagement Exercises, 5-minute presentations, and behaviour questions at the grade's level. Offer to build these once the application is in.
Output format
Unless the user asks otherwise: the evidence map in chat (it's a working tool), then the statement as a document they can edit and export. Show word counts. Keep commentary short; the user is usually close to a deadline.