Plenty of ChatGPT prompt lists for recruiters already run from job description to onboarding survey. What they tend to leave out is an output format, a worked example, and much on bias, consent or data protection. These 55 prompts are written to fill that gap. Each sets a role, asks for your own facts, states what the model must not invent, and fixes the shape of the answer.
The pack covers the whole recruiter job, in-house and agency: the intake meeting and a signed-off brief, adverts that don't quietly screen people out, Boolean and X-ray search, outreach and InMail, screening and volume hiring, structured interviews with anchored scorecards, a CV-to-brief evidence summary, rejections and feedback, offers and pay conversations, and pipeline reporting. Several prompts are for agency work: winning a client, agreeing terms, submitting candidates and chasing feedback.
One rule runs through it: the AI prepares, drafts and analyses, and a person makes every hiring decision. Where bias can actually enter, in ads, requirements, search criteria, screening and interviews, the prompts tell the model to flag proxies for protected characteristics and never to rank people. Four prompts include a worked example with fictional inputs.
Intake & Role Strategy
A search is won or lost in the first meeting with the hiring manager. These prompts help you ask better intake questions, turn messy notes into a signed-off brief, challenge inflated requirement lists, check market expectations against evidence you collect yourself, and plan the timeline backwards from your own stage durations, and, for agency work, agree a job order and terms with a client. They also give you scripts for the conversations most recruiters dread: saying no to the manager without damaging the relationship.
Best for: Preparing a focused intake with a hiring manager
You are an experienced talent partner preparing for an intake meeting with a hiring manager.
Role: {{ROLE_TITLE}}
Team and business context: {{TEAM_CONTEXT}}
What the manager has told me so far: {{MANAGER_NOTES}}
Build a question bank for a 45-minute intake meeting. Group questions under these headings: why the role exists now, what success looks like at 3, 6 and 12 months, the real day-to-day work, must-have skills versus things that can be taught, team dynamics and who the person works with, interview process and decision-makers, compensation and flexibility, and timeline.
Rules:
- Ask about outcomes and skills, not personal traits or backgrounds. If a question invites a proxy for age, gender, nationality, health or family status (for example "digital native", "culture fit", "young and energetic"), do not include it and offer a better question.
- Mark the five questions most likely to change the search strategy with a star.
- Add a short list of "listen for" signals that the brief is unclear or that expectations may be unrealistic.
- Do not recommend candidates or make any hiring decision.
Output: a numbered list under each heading, then a one-paragraph opening script for the first two minutes.
Fill in: ROLE_TITLE TEAM_CONTEXT MANAGER_NOTES
Use it the day before an intake. Print it and fill in the answers as you go.
Best for: Locking down a shared definition of the role
Act as a recruiting operations lead who writes clear role briefs that sourcers, interviewers and the hiring manager can all work from.
Role: {{ROLE_TITLE}}
Raw intake notes (messy is fine): {{INTAKE_NOTES}}
Hiring manager: {{HIRING_MANAGER}}
Turn the notes into a one-page role brief with these sections: purpose of the role, top 5 outcomes in the first year, must-have skills (each with the evidence we would accept), nice-to-have skills, interview stages and who runs them, compensation and location as stated, open questions.
Rules:
- Use only what is in the notes. Where something is missing or contradictory, put it under "Open questions" with a suggested question for {{HIRING_MANAGER}}. Do not fill gaps with assumptions.
- Phrase requirements as skills and outcomes. Flag any requirement that works as a proxy for a protected characteristic, such as a graduation year, "native speaker", or "recent graduate", and propose a neutral wording.
- Keep each section short enough to scan in 30 seconds.
Output: the brief in clear headings and bullets, followed by a list of the three biggest risks to the search.
Fill in: ROLE_TITLE INTAKE_NOTES HIRING_MANAGER
Send the brief back to the hiring manager for sign-off before you start sourcing.
Best for: Cutting inflated requirement lists
You are a calibration coach helping a recruiter test a hiring manager's requirement list. The goal is a requirement list that is honest about what the job needs, so the talent pool is not shrunk for no reason.
Role and level: {{ROLE_TITLE}}, {{SENIORITY}}
Requirement list from the hiring manager: {{REQUIREMENTS_LIST}}
What the person will actually do in their first six months: {{FIRST_SIX_MONTHS}}
For every requirement, tell me:
1. Is it needed on day one, learnable within three months, or only a preference?
2. What job task does it map to? If none is evident, say so.
3. Is it a proxy for something else (years of experience standing in for skill, a named degree or employer standing in for ability, "culture fit")? If so, suggest an assessable alternative.
4. A question I can ask the manager to test it, in a polite and curious tone.
Then give a shortened must-have list of no more than six items and explain what you removed or demoted.
Do not rank people or comment on any individual. Where you are unsure whether a requirement is legally or regulatorily mandatory, flag it for me to confirm instead of guessing.
Output: a table (requirement, category, task it maps to, proxy risk, question), then the shortened list.
Fill in: ROLE_TITLE SENIORITY REQUIREMENTS_LIST FIRST_SIX_MONTHS
Run it before the manager sees a single CV, then use the questions in your next sync.
Best for: Handling difficult hiring-manager conversations
A hiring manager and I are stuck. I need to say something useful without damaging the relationship.
What is going on: {{SITUATION}}
How this manager tends to react: {{MANAGER_STYLE}}
What I can actually point to (pipeline numbers, candidate feedback, declined offers): {{EVIDENCE}}
Give me three ways to raise it. The first leads with data, the second with a question that lets the manager reach the conclusion, and the third gives them two options to choose between. Each should be something I could say in a meeting or paste into a message, under 120 words, opening by acknowledging what the manager is trying to achieve, then using only the evidence above, then making one specific ask, with a fallback if they decline.
If my evidence is thin, say so and tell me what to gather before the conversation. Do not invent figures, quotes or benchmarks. If the manager is asking me to screen people out because of a protected characteristic, say so plainly, suggest I bring in HR or legal, and write a calm, firm line for that moment.
Finish with two lines on when to use which version.
Fill in: SITUATION MANAGER_STYLE EVIDENCE
Good for the awkward conversations: unrealistic pay, endless interview rounds, slow feedback.
Best for: Setting realistic timelines with hiring managers
You are a recruiting project manager. Plan a realistic hiring timeline by working backwards from the date the person is needed.
Target start date: {{START_DATE}}
Our own recent durations for each stage and typical notice period, from the ATS: {{OWN_STAGE_DURATIONS}}
Interview stages and who runs them: {{INTERVIEW_STAGES}}
Known constraints (holidays, panel availability, approvals, background checks): {{CONSTRAINTS}}
Build a week-by-week plan that starts at the start date and runs backwards through: offer accepted, offer approved, final decision, final interviews, first interviews, screening, sourcing and advert go-live, and role approval. For each step give the owner, the output, and the duration, taken from my own figures. Show the date the role must be live to hit the target.
Rules:
- Use only my durations. If a step has no figure, mark it "NEEDS YOUR DATA" instead of estimating.
- Highlight steps where delay is most likely and add one mitigation for each.
- If the target date cannot be met with the constraints given, say so, and show the earliest realistic date and what would need to change to hit the original one.
Output: a table (step, owner, duration, latest date) followed by a short risk list.
Fill in: START_DATE OWN_STAGE_DURATIONS INTERVIEW_STAGES CONSTRAINTS
Share the table with your panel at kickoff so everyone sees their dates.
Best for: Setting up an agency search properly
You are a senior agency account manager preparing a call with a client who wants to give us a vacancy. Help me agree the brief, the way we work and the terms without overpromising.
Client and role: {{CLIENT_NAME}}, {{ROLE_TITLE}}
What the client has told us so far: {{CLIENT_NOTES}}
Our standard terms (fee basis, rebate or guarantee, exclusivity, payment, who can be approached): {{OUR_TERMS}}
What the client's current alternative is (in-house team, other agencies, job boards): {{CLIENT_ALTERNATIVE}}
Create:
1. A list of 12 intake questions that cover the role, success measures, the decision process and who is in it, interview stages, and why previous attempts failed.
2. A short script for explaining our terms in plain words, reading only from the terms above, with a line for each likely challenge (a lower fee, exclusivity, a longer guarantee, "we'll also use other agencies").
3. What to agree before I begin: signed brief, feedback turnaround, candidates already seen, shortlist size.
4. A follow-up email that confirms what was agreed.
Rules:
- Do not state a fee, guarantee, fill time or market fact I have not given you. Mark anything you would need as "CONFIRM WITH MY MANAGER".
- Keep the script respectful and direct, with no pressure. Say where the contract needs legal review.
Output: the four parts as labelled sections.
Fill in: CLIENT_NAME ROLE_TITLE CLIENT_NOTES OUR_TERMS CLIENT_ALTERNATIVE
Agree the commercial terms in writing before you send any candidate.
Inclusive Job Ads & Employer Brand
Adverts are where many qualified people decide not to apply. These prompts write adverts from a real brief, audit existing ones for age-coded and exclusionary wording, turn years-of-experience asks into assessable skills, draft a pay section you can check with HR, and adapt one role to four channels. The employer-brand prompts favour honest day-in-the-life pages and clear process explainers over polish.
Best for: Writing a clear advert for a wide pool
You are a recruitment copywriter who writes job adverts that are clear, honest and welcoming to a wide range of applicants.
Role brief: {{ROLE_BRIEF}}
Company description (2-3 sentences, factual): {{COMPANY_FACTS}}
Salary range and location/work arrangement: {{PAY_AND_LOCATION}}
Tone: {{TONE}}
Write a job advert with these parts: a two-sentence summary, what you will do (5-7 outcome-focused bullets), what you bring (no more than 6 bullets, skills only), what we offer (only benefits the facts support), how we hire (stages and timing), and a brief note that we can adjust the process on request.
Rules:
- Use plain language at about a Year 9 / grade 9 reading level. No jargon, buzzwords or superlatives.
- Avoid gendered, age-coded and exclusionary wording ("ninja", "rockstar", "young", "digital native", "native speaker", "strong man").
- Do not list a requirement the brief does not justify. Do not pad the list with years of experience.
- Do not invent benefits, culture claims or growth promises.
- After the advert, list any words or requirements you changed or want me to check.
Output: the advert, then a short "flagged wording" list.
Fill in: ROLE_BRIEF COMPANY_FACTS PAY_AND_LOCATION TONE
Paste your real brief and facts; the more concrete they are, the less generic the advert.
Best for: Finding exclusionary wording in existing adverts
You are an inclusive hiring reviewer. Audit this job advert for wording and requirements that could discourage qualified people or work as proxies for protected characteristics.
Advert text: {{JOB_AD_TEXT}}
Country or region (relevant for legal terms): {{LOCATION}}
Check for: gender-coded words, age-coded phrases (for example "digital native", "recent graduate", "energetic"), exclusionary language ("native English speaker" where fluency is what is meant), unnecessary degree or employer requirements, unrealistic experience lists, unexplained physical or travel requirements, culture-fit language, and missing information that matters to applicants (pay, location, process, adjustments).
Present your findings as a table with columns: quoted text, issue, why it matters, suggested replacement. Be specific and quote the advert verbatim. If something is a possible legal requirement or bona fide occupational requirement, do not rule on it. Mark it "CONFIRM WITH HR OR LEGAL" instead.
Finish with a revised version of the advert and a one-paragraph summary of the biggest changes. I am responsible for the final text; do not claim the revised version is legally compliant.
Output: the table, the revised advert, the summary.
Fill in: JOB_AD_TEXT LOCATION
Run it on your five most-used templates, not just new ads.
Example output · Claude (example written by The Prompt Index)
Sample inputs (fictional)
- JOB_AD_TEXT: "Sales Coordinator. We're looking for a young, energetic rockstar to join our fast-paced, family-like team. You must be a native English speaker with 10+ years of experience and a degree from a top university. Recent graduates welcome! You'll need to be a strong man to lift stock occasionally. Competitive salary. Apply now, we hire quickly."
- COUNTRY: United Kingdom
Findings
| Quoted text | Issue | Why it matters | Suggested replacement |
|---|---|---|---|
| "young, energetic" | Age-coded | Describes the person, not the work. It can deter older applicants and may be unlawful age discrimination. CONFIRM WITH HR OR LEGAL | "You will manage a busy order flow with frequent changing priorities." |
| "rockstar" | Hype / male-coded jargon | Tells applicants nothing about the job and tends to discourage people who do not see themselves in it. | Delete, or name the outcome: "You will be the main contact for 40 trade customers." (use your real figure) |
| "family-like team" | Culture language | Can signal an expectation of unpaid extra effort and does not describe the work. | "A team of six that meets daily to plan the day's orders." (use your real facts) |
| "native English speaker" | Exclusionary | A nationality or origin proxy. What the job needs is a level of English. CONFIRM WITH HR OR LEGAL | "Fluent written and spoken English for customer calls and emails." |
| "10+ years of experience" | Unjustified, age-correlated; also contradicts "Recent graduates welcome" | Screens out capable people with fewer years and may act as an age proxy. The two statements show the requirement was not thought through. | List the skills: "Experience handling trade orders and resolving delivery queries." |
| "a degree from a top university" | Pedigree filter | Proxy for background and class. Not shown to relate to the work. | "Degree or equivalent practical experience" only if the role really needs a degree. |
| "strong man" | Gendered, and the physical requirement is vague | Excludes women and people with some disabilities. If lifting is a genuine requirement, it should be described. CONFIRM WITH HR OR LEGAL | "Occasionally moves boxes of up to X kg (specify). Reasonable adjustments are available." |
| "Competitive salary" | Missing information | Applicants cannot judge fit without it. Check what your area requires for pay disclosure. | Add the actual range after checking local rules with HR. |
| "Apply now, we hire quickly" | Pressure; missing process | Rushes candidates and says nothing about the stages. | "Two interviews over about two weeks." (use your real process) |
| (missing) | No adjustments statement | Applicants who need adjustments do not know they can ask. | "Tell us if you need any adjustments for the application or interviews." |
Revised advert
Sales Coordinator ([location, hours])
You will be the main contact for our trade customers, taking orders, solving delivery queries and keeping the order log accurate.
What you will do: take and check orders by phone and email; resolve delivery queries with the warehouse; keep customer records up to date; occasionally move boxes of up to [X] kg in the stockroom (adjustments available).
What you bring: experience handling customer orders or similar admin; clear written and spoken English; comfort with spreadsheets; a calm approach when priorities change.
Pay: [range, to be confirmed with HR].
How we hire: two interviews over about two weeks. Tell us if you need any adjustments for the application or the interviews.
Summary of the biggest changes
I removed age-coded, gendered and origin-coded wording, replaced years and pedigree with the skills the work needs, described the lifting requirement honestly, and added pay, process and an adjustments line. I have not verified that the revised text is lawful; please have HR check the items marked.
Best for: Moving from credentials to assessable skills
You are a recruiter who translates vague requirements into assessable ones.
Role: {{ROLE_TITLE}}
Current requirements list: {{CURRENT_REQUIREMENTS}}
The real outcomes expected in the first year: {{FIRST_YEAR_OUTCOMES}}
Rewrite each requirement as a statement of what the person must be able to do, and add how we could assess it fairly (for example a work sample, a structured question or a portfolio review). Replace "X years of experience" and named degrees with the underlying skill unless the role truly needs a regulated qualification, and in that case say "confirm legal requirement".
Rules:
- Keep to a maximum of six must-haves.
- Mark each rewritten item as MUST or HELPFUL.
- Avoid wording that screens by age, background or education prestige.
- Do not invent requirements that are not implied by the outcomes. If an outcome has no matching requirement, point it out.
- Note any requirement you think the hiring manager may push back on, with a one-line reason I can use.
Output: a table with columns (original requirement, rewritten as a skill or outcome, how to assess, MUST or HELPFUL), then a clean bullet list I can paste into an advert.
Fill in: ROLE_TITLE CURRENT_REQUIREMENTS FIRST_YEAR_OUTCOMES
Useful when a manager wants '8+ years' for a role that only needs a specific skill.
Best for: Drafting a transparent pay section
You are a recruitment copywriter drafting the compensation section of a job advert. Pay transparency rules differ by place, so write it carefully and flag what I need to verify.
Role and level: {{ROLE_TITLE}}
Location: {{LOCATION}}
Pay range and structure (salary, bonus, equity if any): {{PAY_DETAILS}}
Benefits confirmed by HR: {{CONFIRMED_BENEFITS}}
Write a short compensation section of no more than 120 words in plain language that states the range clearly, says what the range depends on in general terms (skills and experience relevant to the role), and lists only the confirmed benefits. Add one sentence for how pay is decided at offer stage. Then write a variant for a platform with a character limit of 300.
Rules:
- Use only the facts above. Do not add benefits, rounding or "competitive" claims.
- Do not state what the law requires in {{LOCATION}}. Instead, give me a checklist of the questions to confirm with HR or legal (whether a range is required in the advert, what counts as a good-faith range, what to do with bonus and equity).
- Avoid wording that implies pay will be set by past salary.
Output: the two versions, then the checklist.
Fill in: ROLE_TITLE LOCATION PAY_DETAILS CONFIRMED_BENEFITS
Always confirm the legal checklist with HR; rules change and vary by jurisdiction.
Best for: Consistent messaging across channels
You are a recruitment marketer adapting one role for different channels without changing the facts.
Role: {{ROLE_TITLE}}
Approved facts (what the job is, pay, location, team, process): {{APPROVED_FACTS}}
Ideal reader, described by skills and motivations rather than demographics: {{TARGET_READER}}
Application link: {{APPLY_LINK}}
Write four versions: (1) a full job-board advert of about 250 words, (2) a LinkedIn-style post of under 120 words with a short hook and no hashtag spam, (3) an email newsletter or community-board blurb of three sentences, (4) a message an employee can share with their network, in the first person, that they can edit.
Rules:
- Every version must contain the pay range if it appears in the approved facts.
- Do not add claims that are not in the approved facts.
- Avoid hype, and tell me if my target-reader description targets people by age, gender or another protected trait.
- End with a short list of what to test (hook, headline, length) and how to judge it, using application quality as well as volume.
Output: the four labelled versions, then the test list.
Fill in: ROLE_TITLE APPROVED_FACTS TARGET_READER APPLY_LINK
Publish all four with the same link so you can compare which channel produced good applicants.
Best for: Realistic previews that help candidates self-select
You are an employer-brand writer. Create an honest "day in the life" page for a role from material provided by the team.
Role and team: {{ROLE_AND_TEAM}}
Notes from the team (interviews, quotes, typical week, tools, challenges): {{TEAM_NOTES}}
Things we are NOT good at yet (be candid): {{HONEST_CHALLENGES}}
Write a page of about 350 words with: a plain description of a typical week, who the person works with, what a good first three months looks like, the real challenges, and how decisions get made. Include quotes only if they appear verbatim in the notes, and attribute them to the roles given, not invented names.
Rules:
- No hype, no "family" language, no promises about growth or work-life balance that the notes do not support.
- Include at least two things a candidate might not like, taken from the challenges provided, so the right people self-select.
- Use accessible language, short sentences and no images of any particular type of person implied in the text.
- If the notes are thin, list the questions I should ask the team before publishing.
Output: the page with headings, then a checklist of facts the team should approve.
Fill in: ROLE_AND_TEAM TEAM_NOTES HONEST_CHALLENGES
Have the team lead read it before it goes live; honest pages convert better than glossy ones.
Best for: Reducing candidate drop-off through clarity
You are a candidate-experience writer. Draft a clear explanation of our hiring process so candidates know what to expect.
Role type: {{ROLE_TYPE}}
Stages in order, with who is involved and what happens: {{STAGES}}
Typical timing between stages and when candidates will hear from us: {{TIMING}}
How we handle adjustments and accessibility requests: {{ADJUSTMENTS_POLICY}}
Write a page of 250 to 300 words and a shorter 60-word version for the bottom of job adverts. Include: each stage, length and format, what we assess at each stage (skills and not impressions), what to prepare, how decisions are made, when and how candidates will hear back, and how to ask for adjustments.
Rules:
- State only what the details above say. Do not promise response times, feedback or outcomes that I have not provided.
- Use inclusive language. Do not assume candidates have a particular schedule, device, or background.
- At the end, list the things I must confirm are true before publishing.
Output: the long version, the short version, and the confirm-before-publishing list.
Fill in: ROLE_TYPE STAGES TIMING ADJUSTMENTS_POLICY
Publish it once and link it from every advert and every invitation email.
Sourcing & Boolean Search
Good sourcing finds people by what they can do. These prompts build and debug Boolean strings, design X-ray searches for public portfolios, map adjacent titles and skills, and list target companies to research. Each one tells the model to search on skills and work only, never on proxies for protected characteristics, and one audits your own filters before you use them.
Best for: Writing strings that balance precision and reach
You are a sourcing specialist who writes Boolean search strings for professional networks and applicant databases.
Role: {{ROLE_TITLE}}
Core skills and tools that show up in relevant profiles: {{CORE_SKILLS}}
Location: {{LOCATION}}
Platform and its operator rules (for example, whether it supports parentheses, NOT, quotation marks, wildcards, and its character limit): {{PLATFORM_RULES}}
Build three strings: a narrow one (high precision), a balanced one, and a broad one (high recall). For each string, give:
- the string in a code block, using only operators the platform supports,
- a plain-English explanation of each clause,
- the kind of profile it will miss.
Rules:
- Include title variants and synonyms, grouped in parentheses with OR.
- Search on skills, tools, responsibilities and job titles only. Do not use names, schools, graduation years, age, gender, nationality, photos or any proxy for them, and do not exclude people for career gaps.
- If I gave no platform rules, say your syntax is generic. Never promise a result count.
- Say how the location should be applied: usually through the platform's location filter and radius, not inside the string. Name any clause that may exclude strong candidates and how to loosen it.
Output: the three strings and explanations, then a three-step test plan: run, scan the first 25 results, adjust.
Fill in: ROLE_TITLE CORE_SKILLS LOCATION PLATFORM_RULES
Always test the string yourself and tune it on the first page of results.
Example output · Claude (example written by The Prompt Index)
Sample inputs (fictional)
- ROLE_TITLE: Senior Payroll Analyst
- CORE_SKILLS: UK payroll, PAYE, RTI submissions, pension auto-enrolment, Sage Payroll, Excel, month-end reconciliation
- LOCATION: Leeds, UK (hybrid)
- PLATFORM_RULES: Quotation marks, AND, OR, NOT (uppercase) and parentheses are supported; no wildcards; 2,000-character limit.
Narrow (high precision)
```
("payroll analyst" OR "senior payroll analyst" OR "payroll specialist") AND ("PAYE" OR "RTI") AND ("auto-enrolment" OR "auto enrolment") AND ("Sage" OR "Sage Payroll")
```
- Clause 1 captures title variants. Clause 2 requires UK payroll vocabulary. Clause 3 requires pension auto-enrolment exposure. Clause 4 requires the specific software.
- It will miss: people who use a different system (for example Iris or ADP) and people whose profile lists duties but not the exact terms.
Balanced
```
("payroll analyst" OR "senior payroll analyst" OR "payroll specialist" OR "payroll officer" OR "payroll supervisor") AND ("PAYE" OR "RTI" OR "UK payroll") AND ("auto-enrolment" OR "auto enrolment") AND ("Sage" OR "Iris" OR "ADP" OR "SAP")
```
- Widens titles to adjacent levels, accepts any one of several UK payroll terms, requires pension auto-enrolment exposure, and accepts any common payroll system. "Excel" and "reconciliation" are left out on purpose: nearly every finance profile mentions them, so they filter nothing.
- It will miss: people with sparse profiles and anyone whose payroll work sits under a generalist title such as "finance assistant".
Broad (high recall)
```
("payroll" OR "pay roll") AND ("UK" OR "PAYE" OR "RTI" OR "HMRC") AND ("analyst" OR "specialist" OR "officer" OR "supervisor" OR "manager" OR "administrator")
```
- Casts a wide net on the domain. Expect many results to be more junior or more senior, so the first screen should be quick.
- It will miss: profiles that never use the word payroll.
Location
- Leeds, hybrid: do not put the city in the string. Set the platform's location filter to Leeds with a radius (for example 25 miles, widened if results are thin), because profile location fields and job-title text rarely match.
Flags and loosening
- "Sage" in the narrow string may exclude strong candidates who used another package. Drop it first if the results are thin.
- No clause filters on names, schools, graduation years or career gaps, and none should be added.
- I cannot know how many results any string returns on your platform.
Three-step test plan
1. Run the balanced string and scan the first 25 profiles for fit against the brief.
2. Note which ones you would contact, and which clause let in the ones you would not.
3. Tighten or loosen one clause at a time, re-run, and save the final string with the date in your ATS.
Best for: Finding evidence of skill beyond job boards
You are a sourcing specialist. Design X-ray searches that use a general web search engine to find publicly shared professional work, such as portfolios, open-source contributions, conference talks and published articles.
Role: {{ROLE_TITLE}}
Evidence of skill I want to find (e.g. public code, case studies, talks, writing): {{EVIDENCE_TYPES}}
Sites or domains where this evidence is usually public: {{TARGET_SITES}}
Region or language: {{LOCATION}}
Produce 6 search strings using site: and quotation operators, and for each say what it targets, why it should work, and what the results will not show.
Rules:
- Search for work and skills only. Do not construct searches around personal characteristics, family, health, religion, politics, location of home, or social media.
- Respect each site's terms of use and local privacy and data-protection rules. Tell me to check them, and do not suggest scraping, bypassing logins, or guessing email addresses.
- Remind me that public information may be out of date and that I should contact people through the channel they use for professional contact, with a clear explanation of how I found them.
- Do not tell me that a result means a person is a good fit.
Output: a table (search string, target, rationale, blind spots) and a short note on how to log the source for each person.
Fill in: ROLE_TITLE EVIDENCE_TYPES TARGET_SITES LOCATION
Tell your data-protection lead how you log sourced candidates before running this at scale.
Best for: Opening up a thin talent pool
You are a talent mapping analyst. Help me widen a search by finding the other titles, skills and backgrounds where the same capabilities exist.
Target role: {{ROLE_TITLE}}
Key capabilities the job needs (verbs, not titles): {{CAPABILITIES}}
Industries or company types we already target: {{CURRENT_TARGETS}}
Industries where the capability might exist unseen, if I know any: {{POSSIBLE_ADJACENT}}
Produce:
1. A list of 12 or more alternative job titles for the same work, grouped as "same work, different name" and "adjacent work, needs a bridge".
2. Skills that are transferable, with a note on what a candidate from that background would probably need to learn.
3. Types of organisations where this capability is built, with the reason.
4. Interview questions that would test the transfer fairly.
Rules:
- Say which suggestions you are confident about and which are hypotheses to test. You cannot see live market data, so do not give figures.
- Do not use pedigree (named employers or schools) as a quality signal.
- Note any title that means something different in different countries or sectors.
Output: four headed sections with bullet lists, plus a one-paragraph recommendation of which three adjacent pools to test first and how to measure the test.
Fill in: ROLE_TITLE CAPABILITIES CURRENT_TARGETS POSSIBLE_ADJACENT
Great for roles where the obvious pool is tiny or over-fished.
Best for: Fixing strings that return nothing or everything
You are a sourcing coach. My Boolean search is not returning what I want. Diagnose it.
The string: {{SEARCH_STRING}}
Platform: {{PLATFORM}}
The problem (too few results, too many, wrong profiles, errors): {{PROBLEM}}
Three examples of the kind of person I want to find, described by skills and titles only: {{EXAMPLE_PROFILES}}
Do this:
1. Check the string for syntax issues: brackets that do not match, quotation marks, operator order, uppercase operators, over-long length, and operators that this platform may not support. Say "check the platform's help page" where you are unsure of its rules.
2. Explain what the string is actually asking for, clause by clause.
3. Predict why it fails against the example profiles, using their titles and skills.
4. Give me two corrected versions and describe in one sentence what each one trades off.
Rules:
- Do not add terms that filter on age, gender, nationality, schools, graduation years or similar proxies. If my original string has any, point them out and remove them.
- Do not promise a result count.
Output: numbered diagnosis, the two corrected strings in code blocks, and a quick test routine.
Fill in: SEARCH_STRING PLATFORM PROBLEM EXAMPLE_PROFILES
Paste the exact string, brackets and all. Typos are the most common cause.
Best for: Mapping the market for a hard-to-fill role
You are a market mapping researcher. Help me build a target-company list for a search without relying on prestige.
Role: {{ROLE_TITLE}}
Where the needed skills are built (types of work or products): {{SKILL_ORIGIN}}
Constraints (location, company size, regulated industry, competitors we must not approach for contractual reasons): {{CONSTRAINTS}}
Existing known companies, if any: {{KNOWN_COMPANIES}}
Produce a list of categories of organisation to research, then up to 15 example organisation types or, if you are confident they exist, example company names. For each, give:
- why the skills might be there,
- how to check (what to look for in their products, job ads, engineering or team blogs),
- the risk (non-solicit, conflict, relocation, pay mismatch).
Rules:
- Say plainly when you may be wrong. Company information changes, and you must label every named company as "verify it exists and still does this".
- Do not infer anything about staff from a company's reputation. Never rank companies by the demographics of their staff.
- Do not suggest approaching a competitor's staff in ways that may breach agreements. Tell me to check restrictions with legal.
Output: a table (category or company, why, how to verify, risk) and a ranked order of research with the reasoning.
Fill in: ROLE_TITLE SKILL_ORIGIN CONSTRAINTS KNOWN_COMPANIES
Treat the output as a research to-do list rather than a finished list.
Best for: Auditing your own filters before use
You are an equitable-hiring reviewer. Before I start sourcing or shortlisting, review my search and screening criteria for proxies that could unfairly screen out people.
Role: {{ROLE_TITLE}}
My search or shortlisting criteria (keywords, filters, rules of thumb): {{CRITERIA}}
How I plan to apply them (manually, ATS filter, AI ranking): {{HOW_APPLIED}}
For each criterion, tell me:
- what job skill it is meant to test,
- whether it could correlate with a protected characteristic (age, sex, race, religion, disability, pregnancy, family status, nationality, accent, postcode or school as a stand-in),
- a skills-based alternative.
Look in particular for graduation or career dates, "recent", "senior" or "junior" labels used as age proxies, employment-gap filters, named schools or employers, postcode filters, and language that implies fluency when a role needs only working proficiency.
Rules:
- Do not decide whether a criterion is lawful in my country. Tell me to take it to HR or legal where it is a close call.
- If I plan to use AI to rank or reject candidates, remind me that a person must review the decision and that I should check local rules on automated decision-making.
Output: a table (criterion, intended skill, proxy risk, alternative), then a corrected criteria list.
Fill in: ROLE_TITLE CRITERIA HOW_APPLIED
Do this before you filter, because biased filters are hard to spot after the pool has been shrunk.
Candidate Outreach & Follow-Up
Short, honest, respectful messages get replies. This section covers the first message, a three-step follow-up with a firm stop rule, employee referral requests, re-engaging past finalists, replying to the answers passive candidates give, and an InMail pack, and two agency prompts: a business development email and a speculative candidate pitch. Prompts include a privacy check and keep to professional, public details.
Best for: Writing a respectful first message to a passive candidate
You are a recruiter known for short, respectful messages that people actually reply to.
Role: {{ROLE_TITLE}}
Company and why the role is interesting, in facts: {{ROLE_PITCH}}
What I know about the person's professional work (public, relevant to the role only): {{PROFESSIONAL_DETAILS}}
Pay range and location, if I can share them: {{PAY_AND_LOCATION}}
Channel and length limit: {{CHANNEL}}
Write three versions of the message: one short (under 70 words), one medium (under 110 words), one for email with a subject line. Each should: open with a specific, relevant reason I am contacting them, say what the role is in one line, include the pay range if provided, make one clear low-pressure ask, and offer an easy way to say no.
Rules:
- Mention only professional, public details tied to the role.
- No fake flattery, no urgency tactics, no claims about "perfect fit", and no invented details about their work. If my details are thin, write the message with a plain approach instead of guessing.
- No jargon, no emoji.
- Include a line telling them where I found their details.
Output: the three messages with the subject line, then two ways to personalise each further.
Fill in: ROLE_TITLE ROLE_PITCH PROFESSIONAL_DETAILS PAY_AND_LOCATION CHANNEL
Send fewer, better messages, and log every reply, including 'no thanks'.
Best for: Writing within InMail and connection-note limits
You are a sourcing recruiter writing short outreach for platforms with tight limits: LinkedIn InMail, connection notes and texts.
Role: {{ROLE_TITLE}}
Why it might interest someone, in facts (team, problem, stack, pay band if shareable): {{ROLE_PITCH}}
The person's relevant professional background (public, role-related): {{PROFESSIONAL_DETAILS}}
Platform limits (for example 300-character connection note, 200-word InMail): {{PLATFORM_LIMITS}}
Produce:
- three connection notes within the limit, each with a different opening angle (their work, the problem the team has, a shared community),
- two InMails, one under 60 words and one under 100, each with a subject line under 8 words that does not start with "Opportunity",
- a one-line follow-up for each InMail,
- a plain-text version for SMS or WhatsApp where I already have permission to use that channel.
For each message, put the single clear ask on the last line and make it easy to decline. Give an approximate count for each; I will recount before sending.
If the details I gave are too thin to personalise honestly, write the message with a plain approach instead of guessing at things I have not told you. Never invent a detail about their work.
Output: a labelled list of messages with counts.
Fill in: ROLE_TITLE ROLE_PITCH PROFESSIONAL_DETAILS PLATFORM_LIMITS
Use one angle per batch and compare replies, not just open rates.
Best for: Following up without becoming a pest
You are a recruiter writing a short follow-up sequence for a candidate who has not replied.
Role: {{ROLE_TITLE}}
My first message (paste it): {{FIRST_MESSAGE}}
Facts I can add in later messages (new information, a team update, a clearer pay range, a flexible arrangement): {{NEW_FACTS}}
Channel: {{CHANNEL}}
Write three follow-ups spaced about 4-5 working days apart. Each one must add something new or make the ask easier, such as a different way to engage, rather than repeating "just checking in". The final message should be a graceful close that leaves the door open and offers to stop contacting them.
Rules:
- Keep each under 70 words.
- No guilt, pressure, fake deadlines, or implications that they will miss out.
- Tell me when to stop. After the last message, advise me to respect silence and not contact again unless they respond, and to follow any opt-out or data-protection rules that apply in {{LOCATION}}.
- Suggest what to record in my ATS for each touch.
Output: three messages with timing, then a short note on measuring response without chasing individuals.
Fill in: ROLE_TITLE FIRST_MESSAGE NEW_FACTS CHANNEL LOCATION
Set a firm stop rule. Three polite touches are enough.
Best for: Getting broader, better referrals
You are an internal recruiter who gets good referrals by making it easy and fair.
Role: {{ROLE_TITLE}}
What the role does and who thrives in it, in terms of skills and work: {{ROLE_SUMMARY}}
Referral bonus rules, if any (use the exact wording): {{REFERRAL_TERMS}}
Where to submit referrals: {{SUBMISSION_LINK}}
Write: (1) a short all-staff message of under 150 words, (2) a version for a team lead to personally send to their team, (3) a one-paragraph "how to refer well" guide that explains what information I need and how to talk to a friend or former colleague about the role, (4) a thank-you message for a referrer after the referral is submitted, whatever the outcome.
Rules:
- Ask employees to think about skills and work they have seen, not about people "like us". Include a line explaining that referred people are assessed by the same fair process as everyone else.
- State the bonus terms only as supplied. Do not invent amounts or conditions.
- Warn employees not to share a friend's personal details without their agreement.
- Avoid wording that encourages referring only from existing circles, and encourage looking beyond them.
Output: the four labelled pieces.
Fill in: ROLE_TITLE ROLE_SUMMARY REFERRAL_TERMS SUBMISSION_LINK
Referral networks can replicate a team's makeup, so pair this with at least one other sourcing channel.
Best for: Reactivating past finalists respectfully
You are a recruiter reactivating a candidate who reached a late stage before and was not hired, because a new role may suit them.
Previous role they interviewed for: {{PREVIOUS_ROLE}}
Strengths evidenced in the earlier process, taken from the interview notes (skills and results only): {{EVIDENCED_STRENGTHS}}
New role: {{NEW_ROLE}}
How long ago, and how the last conversation ended: {{LAST_CONTACT}}
Write a short message (under 120 words) and a shorter version for a text or chat platform. Open by acknowledging the previous process in a sentence, say specifically why I am getting back in touch, summarise the new role in two lines, and offer a low-effort next step such as a 15-minute call or simply a "not now".
Rules:
- Use only the strengths from the notes.
- Check that I have permission to contact them: tell me to verify their consent and retention status in the ATS and our privacy policy before sending.
- Do not suggest the earlier outcome was a mistake, and do not promise a different result.
- Provide a one-line version of an "unsubscribe" or "stop contacting me" option.
Output: the two messages and a short pre-send checklist.
Fill in: PREVIOUS_ROLE EVIDENCED_STRENGTHS NEW_ROLE LAST_CONTACT
Check data-retention rules first. Many ATS settings require you to remove or re-consent older candidates.
Best for: Turning replies into conversations
You are a recruiter coach helping me reply to the typical answers I get from outreach.
Role: {{ROLE_TITLE}}
Facts I can share about the role (pay range, work arrangement, team, process): {{ROLE_FACTS_TO_SHARE}}
Facts I must NOT share or don't know yet: {{RESTRICTED_FACTS}}
The reply I received (paste verbatim): {{CANDIDATE_REPLY}}
Classify the reply (interested, curious but cautious, asking about pay, not now, no, asking an off-limits question, confused) and write a response of under 100 words that is warm and direct. Then give two alternative responses with different levels of detail.
Rules:
- Answer questions truthfully with the shareable facts. If a fact is on my restricted list or unknown, say I will find out or that I cannot share it, and do not invent an answer.
- If they say no or not now, accept it, thank them, and ask only whether they would like to be kept in touch on a specific cadence.
- Do not ask about their current salary where that is prohibited or discouraged in my area; tell me to check local rules and ask about expectations instead.
Output: classification, three response options, and the next ATS entry I should make.
Fill in: ROLE_TITLE ROLE_FACTS_TO_SHARE RESTRICTED_FACTS CANDIDATE_REPLY
Keep a library of your best replies, but always match them to what the person actually said.
Best for: Opening a client conversation with a relevant candidate
You are an agency recruiter writing a short email to a client or hiring manager, proposing a candidate before a vacancy exists. This is for agency use and for internal recruiters pitching to their own business.
Client and their known hiring needs or growth signals (facts only): {{CLIENT_SIGNALS}}
Candidate summary, anonymised and with consent to share: {{CANDIDATE_SUMMARY}}
What the candidate wants next (role, location, pay band, availability): {{CANDIDATE_WANTS}}
My agency's terms to reference, if any: {{TERMS_NOTE}}
Write a message of under 150 words, with a subject line, that: states the business problem the candidate could help solve, summarises the candidate's relevant work in three bullet points based only on the summary, and asks for a short call. Add a separate version for LinkedIn of under 60 words.
Rules:
- Share no personal identifiers without the candidate's agreement. Remind me to confirm consent before sending.
- Do not claim the candidate is the best or the only choice, and do not say they will definitely accept.
- Do not guarantee results or timelines.
Output: subject line, both messages, and a consent checklist.
Fill in: CLIENT_SIGNALS CANDIDATE_SUMMARY CANDIDATE_WANTS TERMS_NOTE
Obtain the candidate's explicit agreement before you describe them to anyone.
Best for: Winning a new client without sounding like every other agency
You are an agency recruiter writing a first email to a hiring leader at a company that does not yet work with us.
Prospect's company, role and the hiring signal that prompted this (a job advert, a funding announcement, a team post): {{PROSPECT_SIGNAL}}
What we do well, and one example I can share (anonymised, true): {{OUR_PROOF}}
The problem I think they have and how we would help (facts only): {{HYPOTHESIS}}
My terms to reference, if any: {{TERMS_NOTE}}
Write three emails: a first email under 100 words with subject line, a follow-up two working days later that adds a different useful thing (a market observation I can back up, or a candidate profile summary with consent), and a final polite close. Add a short voicemail script.
Each message should lead with their situation, not our credentials, make one low-effort ask, and avoid pressure and vague claims such as "top talent" or "guaranteed".
Rules:
- Use only the proof and facts I supply; no invented client names, placement numbers or fill times.
- Tell me to check data-protection and marketing-consent rules for cold outreach where the prospect is located.
Output: the three emails and the voicemail.
Fill in: PROSPECT_SIGNAL OUR_PROOF HYPOTHESIS TERMS_NOTE
Never name a past client or a candidate without permission.
Screening & Structured Interviews
Structured, evidence-led process is the most reliable protection against bias you can build. These prompts produce a screening call guide, a structured question set, an anchored scorecard, a work-sample task, an interviewer briefing, cleaned-up notes, an evidence-first debrief, reference checks, a client submission write-up and a question legality check. This section also covers high-volume screening and a CV-to-brief evidence summary. None of them lets the AI make or recommend the hiring decision.
Best for: Consistent, fair first conversations
You are an experienced recruiter designing a 20-minute screening call that gives every candidate the same chance to show their skills.
Role: {{ROLE_TITLE}}
Must-have skills (from the signed-off brief): {{MUST_HAVES}}
Facts I can share about pay, location and process: {{PAY_LOCATION_PROCESS_FACTS}}
Practical questions I need answered (notice, availability, right to work, location): {{PRACTICAL_QUESTIONS}}
Create a call guide with a timed structure: a two-minute opening, five or six questions that test the must-haves, the practical questions, a section where I answer the candidate's questions, and a closing that states the next step and timing. For each skills question, add a "what a strong answer includes" note based on job-relevant evidence.
Rules:
- Ask every candidate the same core questions in the same order.
- Do not ask about age, marital or family status, pregnancy, health, religion, ethnicity, nationality or origin, or about plans to have children. Include a short list of drift topics I should steer away from if a candidate raises them, with a polite phrase to redirect.
- Practical questions must be asked neutrally and only where lawful and needed; tell me to confirm right-to-work wording and pay-history rules in my area.
- Offer to adjust the format on request.
Output: the call guide as a table, then a one-line note template for recording evidence.
Fill in: ROLE_TITLE MUST_HAVES PAY_LOCATION_PROCESS_FACTS PRACTICAL_QUESTIONS
Use the same guide for everyone and write notes against the criteria, not general impressions.
Best for: Screening many applicants quickly without losing consistency
You are a high-volume recruiter. Build a fast, fair screening flow for a role with many hires.
Role and site: {{ROLE_TITLE}} at {{LOCATION}}
Genuine requirements (shifts, right to work, certifications, physical demands, start dates): {{GENUINE_REQUIREMENTS}}
Channel for the first contact (SMS, WhatsApp, a chatbot, phone): {{CHANNEL}}
Number of hires and the timeline: {{HIRING_TARGET}}
Produce:
1. Five knockout questions, each tied to one genuine requirement, with exact wording and yes/no/not-sure answers. Name the requirement each checks; drop any that is not real.
2. Three short skills or situation questions (one line each) with a plain description of what a good answer includes.
3. A text sequence: the first message, a reminder after a day, a confirmation message, a "can't make it" reply, and a no-show follow-up. Keep each under 40 words and include an opt-out. Only message applicants who have opted in to texts or messaging.
4. A short log format for recording answers against the same questions for every applicant.
Rules:
- Questions must be job-related and identical for everyone. Offer an alternative route for anyone who cannot use the channel or needs an adjustment.
- Check shift, right-to-work and background-check wording, and the consent rules for texting applicants where I hire (TCPA- and PECR-style), with HR or legal before use.
Output: numbered sections with the message text in quotes.
Fill in: ROLE_TITLE LOCATION GENUINE_REQUIREMENTS CHANNEL HIRING_TARGET
Test the flow on three colleagues before sending it to applicants.
Best for: Building a defensible, consistent interview
You are an I-O psychology-informed hiring coach. Build a structured interview question set tied to the competencies the role truly needs.
Role: {{ROLE_TITLE}} at {{SENIORITY}} level
Competencies to test (up to 5, from the brief): {{COMPETENCIES}}
Interview length and number of interviewers: {{INTERVIEW_FORMAT}}
Context candidates may come from (industry changers, returners, early-career): {{CANDIDATE_CONTEXTS}}
For each competency write: one behavioural question, one situational question, two follow-up probes, and a note on what evidence to listen for. Allocate time for each competency and assign each to an interviewer so no question is asked twice.
Rules:
- Every question must be answerable by drawing on work, study, volunteering or life experience, so people with non-linear careers are not penalised. Provide an alternative wording for those without direct experience.
- No hypothetical questions that depend on knowing our internal systems.
- Avoid questions that invite protected information (age, family plans, health, background). Do not include "culture fit" or "would I have a beer with them" criteria.
- Questions are identical for every candidate. Probes can vary only by what the candidate says.
Output: a table per competency, followed by a timed agenda for each interviewer.
Fill in: ROLE_TITLE SENIORITY COMPETENCIES INTERVIEW_FORMAT CANDIDATE_CONTEXTS
Link each question to the scorecard so you score evidence on the same dimensions every time.
Best for: Reducing rater drift between interviewers
You are a talent assessment designer. Create an interview scorecard with behaviourally anchored rating scales so different interviewers rate evidence the same way.
Role: {{ROLE_TITLE}}
Competencies and what each means in this role: {{COMPETENCIES}}
Rating scale to use (for example 1 to 4): {{RATING_SCALE}}
For each competency provide:
- a 2-line definition in plain words,
- anchors for each rating point describing what the evidence looks like (specific behaviours, scale and results, not adjectives),
- a "red flag evidence" note that is about job-related behaviour only,
- a field for the evidence quoted from the candidate.
Rules:
- Interviewers must record evidence first, then the rating.
- Provide no overall "hire or reject" box. End with a field for "evidence summary" and "questions still open". The decision will be made by the hiring team after the debrief, and the scorecard must not suggest otherwise.
- No vague labels like "polish", "presence" or "energy" without behavioural anchors, and avoid gendered or age-coded terms.
- Add a note reminding interviewers to score before reading anyone else's ratings.
Output: a scorecard template in a table that can be pasted into a document, followed by a half-page guide for interviewers on how to use it.
Fill in: ROLE_TITLE COMPETENCIES RATING_SCALE
Calibrate the anchors with your interviewers using one example answer per rating.
Best for: Skills-based assessment that respects candidates
You are an assessment designer. Create a work-sample exercise for a role that is realistic, time-boxed and respectful of candidates' time.
Role: {{ROLE_TITLE}}
Core tasks of the job: {{CORE_TASKS}}
Maximum candidate time (and whether it is paid or in-interview): {{TIME_LIMIT}}
Resources available to candidates (tools, data, internet access): {{RESOURCES}}
Produce: a candidate-facing brief written in plain language, the materials or sample data to include, the criteria used to assess it, a marking guide with anchored levels, and a version for candidates who need adjustments (extra time, alternative format).
Rules:
- The task must mirror real work and test no more than three skills from the brief. It must not need unpaid work that the company would use, such as live client work.
- Keep the time within the limit and state the limit on the brief.
- Marks go to the work, anonymised where feasible. Tell me how to remove names from submissions before marking.
- Explain how candidates will get feedback, and that AI-tool use is allowed or not, as I specify in the resources.
- Do not rate "style", background or personality. Do not include anything that needs equipment a candidate may not have.
Output: the brief, materials list, marking guide, and an adjustments note.
Fill in: ROLE_TITLE CORE_TASKS TIME_LIMIT RESOURCES
Pilot the task on a colleague first and time it honestly.
Best for: Briefing occasional interviewers
You are a head of talent writing a briefing for people who interview for us only occasionally.
Role: {{ROLE_TITLE}}
Interviewer's slot (stage, length, competencies assigned): {{INTERVIEWER_SLOT}}
Process rules (note-taking, scoring timeline, who decides): {{PROCESS_RULES}}
Local rules or policies to reference (equal opportunity policy name, data-protection note): {{POLICIES}}
Write a one-page briefing with: the purpose of the interview in two sentences, the questions and time allocated, how to take notes (verbatim evidence, not opinions), how to score before the debrief, what not to ask, what to do if a candidate volunteers personal information, how to handle adjustments, and how to describe the process and next steps without making promises.
Rules:
- Include a "do not ask" list that covers protected characteristics, family plans, health, origin, age and previous pay where relevant, framed in plain words, and mark that it is general guidance and that HR or legal should confirm the specifics for our country.
- Warn against affinity bias, first impressions, and comparing candidates to the previous holder of the role.
- Tell interviewers they do not decide alone and must not discuss candidates outside the debrief.
Output: the one-page briefing, then a 5-item checklist for the 10 minutes before the interview.
Fill in: ROLE_TITLE INTERVIEWER_SLOT PROCESS_RULES POLICIES
Send it 24 hours ahead, together with the scorecard and the CV.
Best for: Cleaning up notes before a debrief
You are an interview notes editor. Convert my raw interview notes into a structured, evidence-based record and flag any subjective or biased language.
Role and competencies being assessed: {{COMPETENCIES}}
My raw notes (messy is fine): {{RAW_NOTES}}
Questions asked (if I recorded them): {{QUESTIONS_ASKED}}
Produce:
1. A table per competency: what the candidate said or did (quoted or closely paraphrased), the evidence it provides, and what the notes do not cover.
2. A list of "opinion words" in my notes (for example "nice", "not a culture fit", "aggressive", "bubbly", "overqualified", "too junior") with a suggestion on how to restate each as observable behaviour or to delete it.
3. A list of any information that was recorded but is not job-related or may be protected (family, health, age, accent) which should be removed from the record, with a note to check our retention policy.
4. Gaps: questions I still need to ask.
Rules:
- Do not add facts that are not in my notes or infer traits.
- Do not provide a recommendation to hire or reject, a score or a ranking. I will score using the scorecard.
Output: the structured notes, the language flags, and the gaps list.
Fill in: COMPETENCIES RAW_NOTES QUESTIONS_ASKED
Run it straight after the interview while you can still fix gaps in your notes.
Best for: Running calibrated hiring decisions
You are a facilitator running a hiring debrief so that evidence leads and the loudest voice does not.
Role: {{ROLE_TITLE}}
Number of candidates and interviewers: {{PARTICIPANTS}}
Scorecard competencies: {{COMPETENCIES}}
Decision owner and how the decision is made: {{DECISION_PROCESS}}
Create a 45-minute debrief agenda and facilitation script. It should include: independent scores submitted in advance, a rule that scores are revealed before discussion, a competency-by-competency review where each interviewer cites evidence, a section for open questions that need follow-up, and a close that records the decision owner's decision and the reasons in evidence terms.
Rules:
- Provide phrases I can use to redirect comments like "I just liked her", "not a culture fit" or "seems too old for it" back to evidence, and to stop discussions of protected characteristics. If one occurs, I should note it and consult HR.
- The AI must not recommend a candidate or a ranking. The decision rests with the people named above.
- Include a short template for recording the decision and the evidence used, and a note on how long to retain it, to be confirmed with our policy.
Output: the timed agenda, the facilitation phrases, and the decision-record template.
Fill in: ROLE_TITLE PARTICIPANTS COMPETENCIES DECISION_PROCESS
Collect scorecards before the meeting and keep the record, because it protects both candidates and you.
Best for: Verifying job-relevant evidence through references
You are a recruiter designing a reference check that is useful, fair and lawful.
Role: {{ROLE_TITLE}}
Competencies the reference should confirm: {{COMPETENCIES}}
Relationship of the referee to the candidate: {{REFEREE_RELATIONSHIP}}
Any concerns raised in interviews that need factual checking (job-related only): {{OPEN_QUESTIONS}}
Produce:
1. An opening script that explains the purpose, confirms the candidate's consent, and states how notes will be used.
2. Eight questions, each tied to a competency, asking for specific examples rather than opinions.
3. Two follow-up probes for each question that are neutral and non-leading.
4. A closing script.
5. A note template recording what the referee said and the exact words used.
Rules:
- Do not ask about health, disability, family, age, religion, origin, union membership or other protected topics. If a referee volunteers them, I should not record them. Add a phrase to politely stop the referee.
- Do not ask for a recommendation to hire or not.
- Remind me to check local rules on references, defamation risk and what the referee's employer allows them to share, and to ask HR if unsure.
- Treat a missing reference as missing information, not negative evidence.
Output: scripts, question table, note template.
Fill in: ROLE_TITLE COMPETENCIES REFEREE_RELATIONSHIP OPEN_QUESTIONS
Always obtain the candidate's agreement before calling anyone.
Best for: Writing a fair, evidence-led submission
You are an agency recruiter writing a submission summary to send a hiring manager alongside a candidate's CV.
Role: {{ROLE_TITLE}}
Client: {{CLIENT_NAME}}
Client's must-have criteria: {{CLIENT_CRITERIA}}
My interview notes on the candidate (skills, achievements, motivations, availability): {{CANDIDATE_NOTES}}
Candidate's expectations (pay, notice, location): {{EXPECTATIONS}}
Write a one-page submission with: a three-sentence summary, a table matching each client criterion to the evidence in my notes (with "not evidenced" where the notes do not cover it), career highlights relevant to the role, what the candidate is looking for, practical details (notice, pay expectations), and questions the client may wish to explore.
Rules:
- Use only my notes. Do not add achievements, exaggerate or guess. Mark gaps clearly.
- Do not include age, date of birth, marital status, family plans, photo, health, nationality or religion, even if present in my notes. Tell me which items you left out.
- Do not recommend whether to hire. Describe fit to the criteria in neutral terms.
- Confirm that the candidate has agreed to the submission and to the disclosure of their expectations.
Output: the submission page, then a list of omitted items and gaps to resolve before sending.
Fill in: ROLE_TITLE CLIENT_NAME CLIENT_CRITERIA CANDIDATE_NOTES EXPECTATIONS
Share the same standard write-up for every candidate you submit.
Best for: Getting a consistent, evidence-only read of a CV
You are a recruiting assistant. Compare one CV with a role brief and report the evidence for each criterion. You are not scoring, ranking or recommending, because the shortlist decision stays with me.
Role brief with must-have and nice-to-have criteria: {{ROLE_BRIEF}}
CV text (remove name, contact details, photo captions and date of birth before pasting): {{CV_TEXT}}
For each criterion in the brief, return a table row with: the criterion; the evidence in the CV, quoted or closely paraphrased with the job or section it comes from; and a status of "Evidenced", "Partly evidenced" or "Not evidenced in this CV". Not evidenced means the CV is silent, not that the person lacks the skill.
After the table, list three questions to fill the biggest gaps and any vague claims (for example "led a team" with no size). List anything non-job-related you saw, such as dates or school names, so I can check you did not weigh it.
Rules:
- Judge against the brief only. Do not infer age, gender, ethnicity, health or family status, and do not treat employment gaps, job titles or employer names as evidence of ability.
- Do not give a score, rating, rank or "good fit / poor fit".
Output: the table, the three questions, the vague-claims list.
Fill in: ROLE_BRIEF CV_TEXT
Run it one CV at a time and decide yourself. Use the same brief for every candidate, and do not add up the status counts into a score.
Best for: Preventing risky questions before they are asked
You are an inclusive hiring reviewer. Check a list of interview questions for bias risk and for questions that may touch protected characteristics.
Role: {{ROLE_TITLE}}
Country or region: {{LOCATION}}
List of questions (including informal ones people like to ask): {{QUESTION_LIST}}
What the question is meant to find out: {{INTENDED_PURPOSE}}
For each question, assess:
- whether it could draw out protected information (age, sex, family plans, health, religion, origin, accent, nationality, marital status, disability, sexual orientation),
- whether it is job-related and necessary,
- whether it could disadvantage people with non-linear careers or caring responsibilities,
- a neutral alternative that tests the same skill.
Present it as a table and add a traffic-light rating: green (fine), amber (reword), red (remove). Then supply a clean list of final questions.
Rules:
- Do not state what is lawful in my country as fact. Add "CONFIRM WITH HR OR LEGAL" to any amber or red item and any question about work authorisation, travel, shifts or physical demands.
- Do not add questions that profile personality, "energy" or "culture fit".
- Treat common questions such as "where are you really from?" and "do you plan to have children?" as red.
Output: the table and the clean list.
Fill in: ROLE_TITLE LOCATION QUESTION_LIST INTENDED_PURPOSE
Review the questions your interviewers actually like to ask, not only the official ones.
Candidate Communications & Rejections
Candidates remember how they were treated long after they forget the job. These prompts write interview invitations with prep and adjustments information, application rejections, evidence-based feedback calls, honest delay updates, replies to adjustment requests, and a client feedback chase for agency work. Each is built to be truthful, avoid empty promises and keep personal details out.
Best for: Making interviews easier to attend and prepare for
You are a recruiter writing a candidate-friendly interview invitation.
Candidate first name: {{CANDIDATE_NAME}}
Role: {{ROLE_TITLE}}
Interview details (date options, length, format, location or link, interviewers' names and roles): {{INTERVIEW_DETAILS}}
What the candidate should prepare or bring, and how to request adjustments: {{PREP_AND_ADJUSTMENTS}}
Write the email with a clear subject line. Include: what the interview is for, the format and timing, who they will meet and what each person does, anything to prepare (and what not to prepare), how to reschedule, how to request adjustments (for example extra time, a quiet space, step-free access, an alternative format) without having to explain why, and when they will hear back.
Rules:
- Under 200 words, plain language, warm and professional.
- Do not promise outcomes or timelines I have not supplied.
- Address the candidate by first name.
- Provide a shorter text-message version and a reminder email to send 24 hours before.
Output: the invitation, the text version, the reminder.
Fill in: CANDIDATE_NAME ROLE_TITLE INTERVIEW_DETAILS PREP_AND_ADJUSTMENTS
Standardise this and send it for every interview so no one is left guessing.
Best for: Rejecting at scale without being cold
You are a candidate-experience specialist. Write a respectful rejection message for people who applied but are not being taken forward at the screening stage.
Role: {{ROLE_TITLE}}
Company name and sign-off: {{COMPANY_AND_SENDER}}
The reason category I can honestly share (for example, we have moved forward with candidates whose experience was closer to the core requirement): {{SHAREABLE_REASON}}
Whether we will keep their details and for how long: {{DATA_RETENTION_NOTE}}
Write three versions: a standard email of 80 to 100 words, a shorter version for an ATS template, and a version for someone who applied to this company more than once. Each should thank them for applying, be clear and kind that they are not progressing, give the shareable reason in general terms, state what happens to their details, and invite them to apply for future roles that match if appropriate.
Rules:
- Be truthful. Do not say "we will keep your CV on file" unless the retention note says so.
- No "overqualified" or "not a fit for our culture" style reasons.
- Do not promise feedback, and do not invite replies you will not answer.
- Avoid corporate clichés and passive phrasing.
Output: the three versions, plus a note on when to send it so candidates are not left waiting.
Fill in: ROLE_TITLE COMPANY_AND_SENDER SHAREABLE_REASON DATA_RETENTION_NOTE
Send rejections promptly. Silence is what candidates remember most.
Best for: Delivering specific feedback you can stand behind
You are a recruiter writing a useful, kind and safe rejection with feedback for a candidate who was interviewed.
Role: {{ROLE_TITLE}}
Candidate's first name: {{CANDIDATE_NAME}}
Evidence-based reasons from the scorecard (skills and examples only): {{SCORECARD_REASONS}}
Strengths worth naming from the interviews: {{STRENGTHS}}
Company policy on feedback (full, limited, none): {{FEEDBACK_POLICY}}
Write a phone script (about 90 seconds) and a follow-up email (about 150 words). Both should thank the candidate, give the decision first, name two strengths, explain the main reason in terms of the role's requirements, and finish with something specific they can do next if useful. Keep to the feedback policy.
Rules:
- Use only the evidence supplied. Do not invent examples or soften with false encouragement.
- Feedback is about evidence against the role requirements, never impressions, and never a comparison with other candidates.
- Do not promise future roles or a reconsideration.
- Check the feedback against the policy and tell me if the reasons I gave could be risky to share, for example if they are subjective, and propose a safer version. I may need to ask HR for advice.
Output: phone script, email, and a list of any phrases I should double-check.
Fill in: ROLE_TITLE CANDIDATE_NAME SCORECARD_REASONS STRENGTHS FEEDBACK_POLICY
Give feedback tied to the scorecard so it is consistent and defensible.
Example output · Claude (example written by The Prompt Index)
Sample inputs (fictional)
- ROLE_TITLE: Operations Coordinator
- CANDIDATE_NAME: Priya
- SCORECARD_REASONS: Strong on stakeholder communication (clear example of resolving a scheduling conflict between two teams). Limited evidence of managing vendor contracts: the example given was in a supporting role. The process-improvement answer described the change but gave no measure of the result.
- STRENGTHS: Clear, organised communication; good questions about how the team works.
- FEEDBACK_POLICY: Limited feedback (two strengths, one development area, tied to the role requirements).
Phone script (about 90 seconds)
"Hi Priya, thanks for taking my call. I wanted to speak to you directly. We've decided to move forward with another candidate for the Operations Coordinator role, and I'm sorry to have to give you that news.
I'd like to share some feedback. The team valued how clearly you communicated, particularly the example where you resolved a scheduling conflict between two teams, and the thoughtful questions you asked about how we work. The main reason for our decision is that the role involves managing vendor contracts directly, and in your examples that work was in a supporting role.
If it helps, one practical step is to build direct vendor-contract experience, for example by asking to lead a supplier renewal in your current role, since that is the gap that decided this one. I appreciate the time you gave us and I'm happy to answer anything about the process itself. I can't promise future roles, but please keep an eye on our careers page."
Follow-up email (about 150 words)
Subject: Your application for Operations Coordinator
Dear Priya,
Thank you for speaking with me today and for the time you put into the interviews.
As discussed, we have decided to continue with another candidate for this role.
Two strengths stood out to the panel: your clear, organised communication, especially your example of resolving a scheduling conflict between two teams, and the thoughtful questions you asked about how the team works.
The main reason for the decision relates to the role's requirement to manage vendor contracts directly. Your examples in this area came from a supporting position.
One practical next step: look for a chance to lead a supplier or vendor renewal in your current role, since direct contract ownership is the gap that decided this. We are grateful for your interest in us. You are welcome to apply for future roles that match your experience.
Kind regards,
[Name]
Phrases to double-check
- "the panel noted" is accurate only if more than one panelist recorded the point; confirm against the scorecards.
- "You are welcome to apply" is an invitation, not a promise. Keep it if that is true of your policy.
- The process-improvement point from the scorecard was left out on purpose because policy allows one development area.
- The suggested next step follows from the vendor-contract gap and is a suggestion, not a promise about future roles.
- Both messages use only the evidence provided and make no comparison to other candidates.
Best for: Keeping candidates informed when things slow
You are a candidate-experience lead writing an honest update when a hiring process has slowed down.
Role: {{ROLE_TITLE}}
Stage the candidate is at and when we last contacted them: {{STAGE_AND_LAST_CONTACT}}
Real reason for the delay that I can share (for example panel availability, budget approval, extra interviews): {{DELAY_REASON}}
What I can commit to and by when, if anything: {{COMMITMENT}}
Write three messages: an email of under 120 words, a text message of under 40 words, and a script for a call. Each should acknowledge the wait, explain in one sentence, say what I can commit to, and give the candidate an easy way to ask questions or step out of the process.
Rules:
- Do not invent a reason, a date or reassurance. If I cannot commit to a date, say so and when the next update will be.
- Do not say or imply "you are the front-runner" or suggest anything about their standing.
- Do not use hype or filler.
- If the delay is likely to exceed a month, suggest wording that honestly lets the candidate decide whether to wait.
- Include a line I can use if the candidate says they have another offer, with the wording to be accurate and without pressure tactics.
Output: the three messages and the competing-offer line.
Fill in: ROLE_TITLE STAGE_AND_LAST_CONTACT DELAY_REASON COMMITMENT
Silence breeds drop-out. An honest update every week or two keeps people informed.
Best for: Keeping a slow client process moving
You are an agency recruiter who needs a client to give feedback on submitted candidates. Slow feedback is losing us good people, and I want to chase without nagging.
Role and client contact: {{ROLE_TITLE}} at {{CLIENT_NAME}}
Candidates submitted, each with date submitted and current status (use anonymous IDs): {{SUBMISSION_LOG}}
What the candidates have been told, and what other processes they are in: {{CANDIDATE_STATUS}}
The feedback turnaround we agreed with the client: {{AGREED_TURNAROUND}}
Write a first chase email under 90 words naming the candidates by ID and asking for the next step on each, a second email three days later offering two easy ways to respond, and a 60-second call script. Then write a holding message to the candidate (under 60 words) that is honest about the delay and does not blame the client.
Rules:
- State the agreed turnaround accurately, without threats. Make it easy for the client to answer in under two minutes.
- Do not tell candidates what the client thinks unless the client has said it.
- Do not invent the client's feedback, a date or a promise.
Output: the two emails, the call script, the candidate message.
Fill in: ROLE_TITLE CLIENT_NAME SUBMISSION_LOG CANDIDATE_STATUS AGREED_TURNAROUND
Chase the same day you reach the turnaround you agreed, not a week later.
Best for: Replying to adjustment requests with care
You are a candidate-experience specialist writing replies to candidates who ask for adjustments in the hiring process.
Role: {{ROLE_TITLE}}
Stage: {{STAGE}}
The request, in the candidate's own words: {{CANDIDATE_REQUEST}}
Adjustments we can usually offer (extra time, alternative format, remote option, sign-language interpreter, step-free access): {{AVAILABLE_ADJUSTMENTS}}
Who to route requests to (HR contact or accessibility lead): {{ROUTING_CONTACT}}
Write a reply of under 150 words that thanks them, confirms we will make adjustments, says exactly what we can do or what we need to check, gives a named contact, and says when they will hear back. Add a chat version.
Rules:
- Do not ask the candidate to disclose a diagnosis, condition or medical evidence. Ask only what they need to participate, and who they would prefer to speak to.
- Do not comment on the candidate's condition or suggest the request affects their chances.
- Do not promise an adjustment that is not on the available list. If the request is outside it, say we will look into it and escalate to the named contact.
- Keep the request confidential: colleagues are told the adjustment only, never the reason.
- Tell me to check my legal duties with HR or legal if the request is unusual.
Output: both replies and an internal handling note.
Fill in: ROLE_TITLE STAGE CANDIDATE_REQUEST AVAILABLE_ADJUSTMENTS ROUTING_CONTACT
Act quickly and keep the details of the request confidential.
Offers & Closing
The last mile is where good searches fail. Use these prompts to prepare the verbal offer call, send an accurate offer cover email, prepare for a counteroffer without pressure tactics, build a compensation rationale that supports consistent pay, and hold an early salary-expectations conversation. The pay prompts use your range and your internal data only.
Best for: Delivering a clear, honest verbal offer
I'm about to make a verbal offer and want to sound like a person, not a script.
What they told us they care about: {{CANDIDATE_PRIORITIES}}
What the team saw in the process that made them the choice (specific examples from scorecards): {{PROCESS_STRENGTHS}}
The offer (salary, bonus, start date, benefits, conditions): {{OFFER_DETAILS}}
Why the pay sits where it does, in internal-equity terms: {{PAY_RATIONALE}}
What can move and what cannot: {{NEGOTIABLE_ITEMS}}
Write what I would actually say, in the order I would say it: a friendly opening that gets to the point inside ten seconds; the offer, every component, stated once, clearly; two sentences on why the team wants them, drawn only from the process strengths I gave you; the conditions and the decision timeline (if there is a real deadline, say why it exists).
Then give short answers to the reactions I'm likely to get: silence, "I need to think", "that's lower than I hoped" and "can you do better?", saying no without apology or pressure where something cannot move.
Mention only what is in the offer, promise nothing verbally about future raises, and say the written offer follows and is the document that counts. Remind me to check the offer against peers in the same role.
End with a three-line follow-up text I can send straight after the call.
Fill in: CANDIDATE_PRIORITIES PROCESS_STRENGTHS OFFER_DETAILS PAY_RATIONALE NEGOTIABLE_ITEMS
Practise the first ninety seconds out loud, particularly the pay figure.
Best for: Handling pay expectations honestly and early
You are a recruiter coach. Help me hold an honest conversation with a candidate about pay, early in the process, so neither side wastes time.
Role and approved pay range: {{ROLE_TITLE}}, {{PAY_RANGE}}
Whether I may share the range, and how much of the package I can describe: {{SHARING_RULES}}
What the candidate has said about their expectations so far: {{CANDIDATE_SIGNALS}}
Where I think there may be a gap: {{POSSIBLE_GAP}}
Write a natural opener for the pay conversation; a neutral question to ask about their expectations (not their current or past pay, because rules on salary history differ by place and I will confirm mine); a response if the expectation is inside the range, a response if it is above, a response if they ask me to go first, and a response if they decline to say. Then give a short note I can send afterwards that records what was discussed.
If the expectation is above the range, help me say so respectfully and ask what else in the package matters, without implying the range will move.
Do not invent market data. List the two facts I should check with HR before the call (what I may share and what I may ask).
Output: the labelled scripts and the follow-up note.
Fill in: ROLE_TITLE PAY_RANGE SHARING_RULES CANDIDATE_SIGNALS POSSIBLE_GAP
Have this conversation early. A pay gap found late costs everyone time.
Best for: Sending an accurate, welcoming offer email
You are a recruiter writing the email that accompanies a formal written offer. It is not the legal contract.
Candidate first name: {{CANDIDATE_NAME}}
Role: {{ROLE_TITLE}}
Start date: {{START_DATE}}
Key terms, summarised from the offer document (pay, location, notice, conditions): {{KEY_TERMS}}
Decision deadline and the contact for questions: {{DEADLINE_AND_CONTACT}}
Next steps after acceptance (paperwork, checks, onboarding): {{NEXT_STEPS}}
Write a warm and clear email of 150 to 180 words with a subject line. Include a congratulation, a short statement of why the team is excited using only facts supplied, a summary of the key terms with a note that the attached document governs, the deadline and what happens next.
Rules:
- Do not contradict, expand or add to the offer terms. If something in the key terms looks incomplete or inconsistent (for example, no stated notice or conditions), list it for me to check before sending.
- No promises about future pay, promotion or flexibility unless in the supplied terms.
- Do not mention other candidates.
- Avoid pressure language.
Output: the email and a pre-send checklist (document attached, figures match, approvals obtained).
Fill in: CANDIDATE_NAME ROLE_TITLE START_DATE KEY_TERMS DEADLINE_AND_CONTACT NEXT_STEPS
Read the figures against the signed offer document, not from memory.
Best for: Handling counteroffers without pressure
A candidate who has accepted, or is close to accepting, may get a counteroffer from their current employer. I want to handle it honestly and without pressure. Be my sounding board.
What they told us pushed them to look: {{CANDIDATE_REASONS}}
Our offer against what they said they wanted: {{OFFER_VS_WANTS}}
What we can still adjust (through the hiring manager and HR): {{FLEXIBILITY}}
First, list the open, curious questions I should ask now to find out what would make staying or leaving the right move, so that a counteroffer is not a surprise. Second, give me honest talking points on how our role relates to the reasons they gave. No pressure, no exaggeration, and no claims about their current employer. Third, write two or three lines I can use if they say "they've matched it". Fourth, write down what I should never say: threats, false deadlines, promises I cannot keep. Fifth, tell me how to take a "no" gracefully and keep the relationship.
Throughout, remember the decision is theirs. If a pay change would help, remind me to route it through the hiring manager and HR so it stays consistent with peers.
Format: five short headed sections, scripts in quotation marks.
Fill in: CANDIDATE_REASONS OFFER_VS_WANTS FLEXIBILITY
The goal is a decision the candidate will still be happy with in six months.
Best for: Setting offers consistently and explaining them
You are a compensation-aware recruiter assembling a brief for the hiring manager and HR so an offer is fair and defensible.
Role: {{ROLE_TITLE}}
Level: {{SENIORITY}}
The approved pay range: {{PAY_RANGE}}
Candidate's relevant skills and evidence from the process (job-related only): {{CANDIDATE_EVIDENCE}}
Pay of current employees in similar roles (anonymised, internal): {{INTERNAL_PEERS}}
Produce a one-page pack that sets out: where in the range the offer should sit and why, using job-related criteria such as skills against the competencies, scope and market evidence I will supply; how it compares with current peers; any equity concerns to resolve first; and the questions for HR to confirm.
Rules:
- Base the recommendation on the documented criteria and the internal range only. Do not use the candidate's pay history, negotiation style, gender, age, ethnicity or perceived "need".
- Highlight if offering above peers in the same role could create an internal pay inequity, and if so, show options.
- Do not invent market figures. Where market data would help, state "TO CHECK" with a suggested source.
- You are producing input for people who decide; do not state a final offer as the decision.
- Mention that pay-equity and pay-transparency rules vary and that HR should review.
Output: the one-page pack and a list of questions for HR.
Fill in: ROLE_TITLE SENIORITY PAY_RANGE CANDIDATE_EVIDENCE INTERNAL_PEERS
Keep a record of the reasoning so you can explain any offer later.
Pipeline Reporting & Process
Numbers help you argue for change when they are honest. These prompts turn an ATS export into a weekly report, diagnose funnel drop-off, compare sourcing channels on cost and quality, prepare a privacy-safe funnel fairness review for HR and legal, and shape a budget case for finance. They calculate only from your data, show the arithmetic, and refuse to invent benchmarks.
Best for: Updating managers quickly from ATS numbers
You are a recruiting operations analyst who writes weekly updates that hiring managers actually read.
Data: pipeline export for the week (paste the table; remove names and personal data first): {{PIPELINE_DATA}}
Roles covered and target start dates: {{ROLES_AND_TARGETS}}
Notes on context (holidays, freezes, changes in scope): {{CONTEXT_NOTES}}
Write a one-page report with: a three-sentence headline, a table per role (stage, count, change from last week, next action and owner), blockers that need a decision from someone, and risks to the target date. Calculate only from the numbers I provide, and show the arithmetic.
Rules:
- If any number looks inconsistent (for example, more interviews than screened candidates), flag it instead of silently correcting it.
- Do not predict hiring outcomes or offer acceptance, and do not invent benchmarks or targets. Where a comparison would help, tell me what data to add.
- Refer to candidates by anonymous ID only.
- Keep it neutral. Describe what the data shows, then suggest options. Leave decisions to the people who own them.
Output: the report, then a list of questions the data raises that I should check.
Fill in: PIPELINE_DATA ROLES_AND_TARGETS CONTEXT_NOTES
Strip names and personal data from the export before pasting it into any AI tool.
Best for: Finding the stage that loses people
You are a recruiting analyst. Help me work out where and why candidates leave the process, using only the data I provide.
Role family: {{ROLE_FAMILY}}
Stage counts and elapsed days between stages (aggregate numbers, no personal data): {{STAGE_DATA}}
Reasons given for withdrawal or decline, where recorded: {{WITHDRAWAL_REASONS}}
Changes made recently to the process: {{PROCESS_CHANGES}}
Do this:
1. Calculate stage-to-stage conversion rates and the time at each stage. Show the maths.
2. Identify the two or three biggest drop-offs or delays, and say which are large enough to be meaningful and which may be noise given the sample size.
3. Give hypotheses for each, such as unclear pay, long gaps between steps, interview load, communication quality, with the evidence needed to test each.
4. Propose one small experiment per hypothesis and how I would measure it.
Rules:
- Do not compare my numbers with external benchmarks or claim a conversion is "good" or "bad" unless I give you a benchmark and its source.
- Correlation is not cause, so state that.
- Flag where the sample is under about 30 and where reasons for withdrawal are missing.
- Do not suggest rushing candidates or cutting assessments in ways that reduce fairness.
Output: calculations table, findings, hypotheses and experiments.
Fill in: ROLE_FAMILY STAGE_DATA WITHDRAWAL_REASONS PROCESS_CHANGES
Look at drop-off by stage and by wait time between stages before blaming candidates.
Best for: Spending the sourcing budget on what works
You are a talent acquisition analyst. Evaluate which sourcing channels deliver people we hire and keep, not just the most applications.
Channels used: {{CHANNELS}}
For each channel: applications, screened, interviewed, offers, hires, cost where known, and early retention data where available: {{CHANNEL_DATA}}
Period covered and any unusual events: {{PERIOD_AND_EVENTS}}
Produce a table of conversion from application to hire per channel and, where data is supplied, cost per hire and early retention. Then rank channels on cost and on quality separately, because they may differ, and explain the limitations of small samples.
Rules:
- Calculate only from the data. Show formulas. If cost or retention data is missing, say that the comparison cannot be made rather than guessing.
- Do not state industry averages or benchmarks.
- Note that channels draw different candidate pools. Tell me to look at whether any channel is under-represented in the final stages, and to review that with HR using aggregated, privacy-safe data.
- Recommend actions such as testing a channel, stopping one, or collecting more data, with the evidence for each.
- Do not recommend stopping a channel based on fewer than 10 hires.
Output: tables, a limitations list, and three recommended actions.
Fill in: CHANNELS CHANNEL_DATA PERIOD_AND_EVENTS
Combine channel data with hiring-manager satisfaction where you can.
Best for: Preparing a careful, privacy-safe funnel review
You are an analyst supporting an HR-led review of whether our hiring funnel affects groups differently. You help organise the analysis; HR and legal interpret it.
Aggregated counts at each stage by self-reported group, collected voluntarily and separately from selection decisions: {{AGGREGATED_COUNTS}}
Stages in the funnel and how decisions are made at each: {{STAGES_AND_DECISIONS}}
Local requirements for monitoring (e.g. adverse impact rules), if known: {{LOCAL_REQUIREMENTS}}
Do this:
1. Calculate pass-through rates by stage and group, and the ratio of each group's rate to the highest group's rate (one common method; confirm which applies where you hire). Show the arithmetic.
2. Flag cells with small counts (under about 10) as unreliable and do not draw conclusions from them.
3. List possible explanations to investigate (sourcing channels, assessment design, rater effects) and the evidence to test each.
4. List the questions HR and legal should answer, including how the thresholds in my jurisdiction apply.
Rules:
- Do not conclude that discrimination did or did not occur, and do not state a legal threshold as fact. Say where I must check.
- Do not infer anyone's characteristics from names or other data.
- Keep individual candidates out of the analysis.
- Recommend that conclusions and actions are approved by HR or legal.
Output: calculation tables, a reliability note and the questions list.
Fill in: AGGREGATED_COUNTS STAGES_AND_DECISIONS LOCAL_REQUIREMENTS
Only use counts collected lawfully for monitoring, and keep that data separate from hiring decisions.
Example output · Claude (example written by The Prompt Index)
Sample inputs (fictional; groups are anonymous labels)
- AGGREGATED_COUNTS: Group A / B / C. Applied 200 / 120 / 40. Screened 60 / 30 / 8. Interviewed 24 / 10 / 3. Offered 8 / 3 / 1.
- STAGES_AND_DECISIONS: Screen = recruiter CV review; interview = panel with scorecard; offer = hiring manager with HR.
- LOCAL_REQUIREMENTS: Not known.
1. Pass-through rates (count at next stage / count at this stage), with the ratio to the highest group (one common method; confirm which applies where you hire)
| Stage | Group A | Group B | Group C |
|---|---|---|---|
| Applied to screened | 60/200 = 30.0% | 30/120 = 25.0% | 8/40 = 20.0% |
| Screened to interviewed | 24/60 = 40.0% | 10/30 = 33.3% | 3/8 = 37.5% |
| Interviewed to offered | 8/24 = 33.3% | 3/10 = 30.0% | 1/3 = 33.3% |
Ratio of each group's rate to the highest group's rate
| Stage | A | B | C |
|---|---|---|---|
| Applied to screened | 1.00 | 25.0/30.0 = 0.83 | 20.0/30.0 = 0.67 |
| Screened to interviewed | 1.00 | 33.3/40.0 = 0.83 | 37.5/40.0 = 0.94 |
| Interviewed to offered | 1.00 | 30.0/33.3 = 0.90 | 33.3/33.3 = 1.00 |
2. Reliability
- Group C has 8 screened, 3 interviewed and 1 offered. These counts are too small for any conclusion: one more person interviewed would move Group C's screened-to-interviewed rate by 12.5 points (1/8), and one more offer would move its interviewed-to-offered rate by 33.3 points (1/3).
- Group B's later stages (denominators 30 and 10) are also small: one person moves its rates by 3.3 and 10 points. Treat everything after the screen as indicative at most.
- The largest gap is at the first stage (application to screen). Its denominators are larger (200, 120, 40), but Group C's 8 screened is still below the threshold of about 10, so even this gap is indicative only. It is the one to look at first, not a finding; one person moves Group C's rate by 2.5 points (1/40).
3. Possible explanations to investigate (none confirmed)
- Sourcing mix: do groups arrive through different channels with different applicant quality or role fit? Test by splitting the screen rate by channel.
- Screening criteria: are the CV filters or keywords correlated with group membership? Review the criteria with a proxy check.
- Reviewer effects: do different recruiters screen at different rates? Compare by reviewer, without identifying individuals in the report.
- Applicant volume: if one group applied broadly to unsuitable roles, the screen rate may reflect fit and not the process. Check against the role requirements.
4. Questions for HR and legal
- Which monitoring method and threshold apply in our jurisdiction, and how should small samples be treated?
- Are the group counts collected lawfully, voluntarily and stored separately from decisions?
- What is the minimum sample size before we act, and who approves actions?
- What should we do about the first-stage gap while the numbers are small?
This analysis neither finds nor rules out discrimination. It organises the numbers for people qualified to interpret them.
Best for: Securing budget with an honest, evidence-based case
You are a talent acquisition leader preparing a short business case to a finance partner for recruiting resources (a recruiter, a tool, an agency budget, or an employer-brand project).
Request: {{REQUEST}}
Problem it addresses, with our own data (time to fill, open roles, hiring-manager hours, agency spend): {{OWN_DATA}}
Cost of the request and the period: {{COST}}
Alternatives we considered: {{ALTERNATIVES}}
Write a one-page business case with: the problem in one paragraph, the evidence from our data, the proposal, the cost, the expected effect expressed as a range with assumptions stated, risks, alternatives and why they were not chosen, and how we will measure success at 90 and 180 days.
Rules:
- Use only the numbers I provide. Show every calculation. Where I have no data, say what data to gather instead of estimating.
- Do not invent savings, benchmarks or vendor claims. Any expected effect must be presented as an assumption and not a guarantee.
- Include what would make us stop or change course.
- Keep the language plain for a finance reader, without jargon or recruitment slang.
- Include a short list of the biggest weaknesses in this case so I can prepare for questions.
Output: the one-page case and the weaknesses list.
Fill in: REQUEST OWN_DATA COST ALTERNATIVES
Bring your own data. A short case with real numbers beats a long one without.