By role · Free prompt pack

ChatGPT, Claude & Gemini Prompts for HR Professionals & People Teams

55 free AI prompts for HR generalists: policies, onboarding, reviews, difficult conversations, surveys, metrics and investigations, with legal-risk flags.

Prompts
55
Works with
ChatGPT, Claude & Gemini
Updated
Price
Free to copy

The popular HR prompt lists we read (Leena AI, AIHR and ClickUp) are mostly short requests with no output format, no worked example and little about what to verify under local employment law. This page is for the generalist who runs people operations at a small or mid-size company, often alone, with a handbook to fix, a review cycle to get through and a manager waiting outside the door.

The 55 prompts cover the work that fills an HR week: policies and handbooks, onboarding and offboarding, probation and absence, performance reviews and difficult conversations, employee communications, engagement and fairness checks, training and succession, HR metrics, and the sensitive cases (investigations, accommodations, grievances, warnings, dismissals). Sourcing, job adverts and candidate screening are left to the recruiters pack.

Each prompt names its inputs, its constraints and the output you should get back, and four come with a worked example. Prompts that take case facts tell you to use labels like Employee A, and the sensitive ones tell the model to flag legal risk and list what to verify locally. Employment law differs by country and state, so treat every output as a first draft for a person to check.

They are free to copy and work in ChatGPT, Claude and Gemini.

Policies, Handbooks & Compliance Admin

Drafting a policy is the easy half; getting people to read it is the hard half. These prompts draft and simplify policies, audit a handbook for gaps, compare leave models, write an AI-use policy, and plan the rollout. Policy text that depends on local law is marked for checking rather than stated as fact.

Remote and Hybrid Work Policy Draft

Best for: A first draft of a policy managers can apply consistently

You are an HR generalist at a company of {{COMPANY_SIZE}} people. Draft a remote and hybrid work policy that managers can apply consistently.

Location of employees: {{COUNTRY_OR_STATE}}
Work model: {{WORK_MODEL}}
Non-negotiable rules from leadership: {{KEY_RULES}}
Equipment, expenses and home-office support we offer: {{SUPPORT_OFFERED}}

Cover: purpose and scope; who is eligible and how exceptions are requested; core hours and availability; in-office expectations; equipment and expenses; data security at home; working from another city or country (flag tax, immigration and insurance questions rather than answering them); health and safety; how the policy is reviewed.

Constraints: plain English at a grade-9 reading level, no unexplained jargon, no promises we haven't listed above. Where you had to assume something, mark it [ASSUMPTION]. Where local law decides the answer, mark it [CHECK LOCAL LAW] and say what to check.

Output: a numbered policy with short headings, then a separate list of every [ASSUMPTION] and [CHECK LOCAL LAW] item, then three questions leadership still needs to decide.

Fill in: COMPANY_SIZE COUNTRY_OR_STATE WORK_MODEL KEY_RULES SUPPORT_OFFERED

Use for a first draft, then have your lawyer or employment adviser check the flagged items before it goes out.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Rewrite a Policy in Plain English

Best for: Making a dense policy readable without changing its meaning

You are an HR communications specialist. Rewrite the policy below so a busy employee can understand it on one read, without changing what it means.

Policy text: {{POLICY_TEXT}}
Audience: {{AUDIENCE}}
Country or state: {{COUNTRY_OR_STATE}}

Steps:
1. Identify every obligation, right, deadline, approval step and consequence in the original.
2. Rewrite in short sentences, active voice, and "you" and "we". Replace legal-sounding phrases with everyday ones, unless the exact wording has legal force. If it might, keep it and add a plain-English note beside it.
3. Add a "What this means for you" box of no more than five bullets.
4. List anything in the original that is ambiguous, contradictory or missing, as questions for the policy owner.

Constraints: do not add new rules, remove obligations or soften consequences. If you are unsure whether a change alters the meaning, leave the original wording and flag it. Say which parts should be checked against {{COUNTRY_OR_STATE}} employment law before publishing.

Output: (a) the rewritten policy, (b) the "What this means for you" box, (c) a table with columns Original phrase, Plain version, Meaning changed? (Yes/No/Unsure).

Fill in: POLICY_TEXT AUDIENCE COUNTRY_OR_STATE

Paste one policy at a time. Compare the table against the original before publishing.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Handbook Gap Audit From Your Table of Contents

Best for: Building a handbook refresh to-do list

You are an experienced HR consultant reviewing an employee handbook for a small or mid-size employer. You have only the table of contents, so you are looking for gaps to investigate, not giving a legal opinion.

Company headcount: {{HEADCOUNT}}
Industry: {{INDUSTRY}}
Where employees work: {{COUNTRY_OR_STATE}}
Handbook table of contents: {{HANDBOOK_TOC}}

Do this:
1. List the policy areas a handbook for this kind of employer normally covers, grouped as Essential, Common and Situational (for example remote work, overtime, safety, data protection, social media, AI tool use, whistleblowing, leave).
2. Compare that list with the table of contents and show what is missing, what looks thin, and what may be out of date or duplicated.
3. For each gap, say why it matters in one sentence and how urgent it is (High, Medium, Low).
4. Say which gaps depend on {{COUNTRY_OR_STATE}} law, and label those [VERIFY LOCALLY]. Do not cite specific statutes or section numbers unless you are certain, and then tag them [VERIFY].

Constraints: don't assume a topic is missing just because the heading is worded differently; ask me when unsure.

Output: a table with columns Policy area, Status, Why it matters, Urgency, Needs local check, followed by a suggested order of work for the next 90 days.

Fill in: HEADCOUNT INDUSTRY COUNTRY_OR_STATE HANDBOOK_TOC

Run this before an annual handbook refresh to build the to-do list, not to decide what the law requires.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Time Off and Leave Policy Options Compared

Best for: Comparing PTO models before a policy change

You are a people-operations adviser helping a company choose how to handle time off. Present options neutrally; the decision is ours.

Company size and type: {{COMPANY_PROFILE}}
Where staff are based: {{COUNTRY_OR_STATE}}
Current approach and its problems: {{CURRENT_APPROACH}}
Our priorities (for example fairness, cost, flexibility, cover): {{PRIORITIES}}

Compare three realistic models for paid time off (for example fixed allowance, accrual, flexible/open allowance) plus the option of keeping the current approach.

For each option give: how it works in practice; who it suits and who it disadvantages; cost and admin implications; common ways it goes wrong (for example staff taking too little leave); what managers would need to do differently.

Constraints: statutory minimums, carry-over rules and payout-on-exit rules vary by location. Do not state them as fact. Instead list the questions I must confirm for {{COUNTRY_OR_STATE}}. Don't invent benchmarks or industry statistics.

Output: a comparison table with the options as columns and the factors as rows, a recommendation framed as "if you value X most, lean toward Y", and a short list of things to verify before changing anything.

Fill in: COMPANY_PROFILE COUNTRY_OR_STATE CURRENT_APPROACH PRIORITIES

Use when you're deciding between leave models; bring the table to the leadership conversation.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Acceptable AI Use Policy for Employees

Best for: A usable AI policy for a company with no precedent

You are an HR and information-governance adviser writing an acceptable-use policy for generative AI tools such as ChatGPT, Claude and Gemini.

Our business and data sensitivity: {{BUSINESS_AND_DATA_TYPES}}
Tools currently approved: {{APPROVED_TOOLS}}
Teams that use AI most: {{HEAVY_USER_TEAMS}}
Country or state: {{COUNTRY_OR_STATE}}

Draft a one-to-two page policy including: purpose; what is allowed; what is never allowed (personal data about employees or customers, confidential client material, passwords, unpublished financials, unless the approved tool's contract permits it); human review of output before it is used externally or in decisions about people; no AI-only decisions about hiring, pay, discipline or dismissal; disclosure of AI use to clients where required; how to report a mistake or data leak; consequences and who to ask.

Constraints: practical, not scary. Use plain language. Give three examples of acceptable use and three of unacceptable use specific to our business. Mark any rule that depends on data protection or employment law as [CHECK WITH LEGAL] and name what to check.

Output: the policy, then a one-paragraph summary for a staff announcement, then five questions for IT and legal to answer before launch.

Fill in: BUSINESS_AND_DATA_TYPES APPROVED_TOOLS HEAVY_USER_TEAMS COUNTRY_OR_STATE

Adapt the examples to your tools, then have IT and legal approve it.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Policy Rollout Plan and Manager FAQ

Best for: Launching a policy change without surprises

You are an HR business partner planning how to launch a new or changed policy so that it is understood, not just published.

Policy summary: {{POLICY_SUMMARY}}
Effective date: {{EFFECTIVE_DATE}}
Who is affected: {{AFFECTED_GROUPS}}
Likely objections or concerns: {{LIKELY_CONCERNS}}

Produce:
1. A rollout timeline working backward from the effective date: leadership briefing, manager briefing, all-staff announcement, reminders, and a review point 60 days after launch.
2. A manager FAQ with 10 questions employees are likely to ask, each with a short, honest answer drawn only from the summary above. Where the summary does not answer the question, write "Not decided: escalate to HR" instead of guessing.
3. Three things managers must NOT say or promise (for example exceptions that haven't been approved).
4. Two ways to check, after launch, whether the policy is being understood and applied consistently.

Constraints: if the change affects pay, hours, contract terms or the number of roles, remind me to check consultation, notice and consent requirements for my location and tell me to ask an employment lawyer. Don't assume the change is lawful.

Output: a timeline table, the FAQ in Q and A format, and a one-line summary of the main risk.

Fill in: POLICY_SUMMARY EFFECTIVE_DATE AFFECTED_GROUPS LIKELY_CONCERNS

Give managers the FAQ before employees see the announcement.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Annual HR Compliance Calendar Starter

Best for: Setting up an annual HR task calendar

You are an HR operations lead building a recurring compliance and admin calendar for a small or mid-size employer.

Location(s): {{COUNTRY_OR_STATE}}
Headcount and types of workers (employees, contractors, seasonal): {{COMPANY_PROFILE}}
Industry: {{INDUSTRY}}
HR systems we use: {{HR_SYSTEMS}}

Build a month-by-month calendar of recurring HR activities: policy reviews, handbook acknowledgements, mandatory training refreshers, performance cycles, benefits renewal and open enrolment, payroll year-end tasks, right-to-work or work-authorisation document checks, safety inspections, survey cycles, and records retention reviews.

Constraints: I know deadlines and filing requirements differ by place. Do NOT give specific legal deadlines or form names unless you are certain; otherwise write "deadline: confirm with [authority or adviser]". Separate "legal or regulatory" items from "good practice" items so I can see which are optional. Assume nothing about our benefits renewal date; ask me, or insert a placeholder.

Output: a table with columns Month, Task, Type (Legal/Regulatory, Contractual, Good practice), Owner (suggested), Notes. Finish with a list of items I must verify before relying on the calendar.

Fill in: COUNTRY_OR_STATE COMPANY_PROFILE INDUSTRY HR_SYSTEMS

Treat it as a scaffold. Confirm every legal date with your adviser or the official source.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Onboarding & Offboarding

From the welcome email to the last-day checklist. These cover pre-boarding, a 30-60-90 plan, remote first weeks, early check-ins, probation reviews, offboarding access, knowledge handover and exit interviews. Final pay, notice and benefits rules differ by location, so the prompts flag them to verify.

Pre-Boarding Welcome Email and Checklist

Best for: Making a new hire's first day run smoothly

You are an HR coordinator at {{COMPANY_NAME}}. A new hire accepted the offer and starts on {{START_DATE}}. Write the pre-boarding email and a checklist that makes their first morning calm.

New hire's role and team: {{ROLE_AND_TEAM}}
Work arrangement (office, hybrid, remote): {{WORK_ARRANGEMENT}}
Our tone: {{COMPANY_TONE}}
Things we already do: {{EXISTING_STEPS}}

Write:
1. A warm email under 200 words that tells them where to be and when, who greets them, what to bring, what happens on day one, and who to contact with questions. Do not invent details I haven't given; use [BRACKETS] for anything missing.
2. A checklist split into "New hire does" (paperwork, documents, forms) and "We do" (equipment, accounts, desk or courier, welcome from manager, buddy assigned).
3. A short list of items to confirm before the email is sent.

Constraints: no attachments of personal data in the email; sensitive forms go through your secure system. Remind me to check document and right-to-work requirements for the location rather than listing them from memory. Keep it friendly but not gushing.

Output: the email, then two checklists in bullet form, then the confirmation list.

Fill in: COMPANY_NAME START_DATE ROLE_AND_TEAM WORK_ARRANGEMENT COMPANY_TONE EXISTING_STEPS

Send about a week before the start date.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

30-60-90 Day Onboarding Plan for a New Hire

Best for: A structured plan a manager can run themselves

You are a people-operations specialist designing an onboarding plan that a new manager can run without extra HR help.

Role: {{ROLE_TITLE}}
Team goals this quarter: {{TEAM_GOALS}}
Manager's style and available time: {{MANAGER_STYLE}}
Tools and systems the person must learn: {{TOOLS_AND_SYSTEMS}}
Key people to meet: {{KEY_PEOPLE}}

Create a 30-60-90 day plan with, for each phase: learning goals, relationships to build, concrete deliverables, check-in questions for the manager, and a "done well looks like" description that can be observed rather than a vague feeling.

Constraints: make the first 30 days mostly about learning and relationships, the second about contributing under guidance, the third about owning something small. Keep the plan realistic for the manager's stated time. Don't invent company processes. If a step depends on something I haven't told you, mark it [CONFIRM]. Avoid jargon.

Output: a table with columns Phase, Goals, People and meetings, Deliverables, Manager check-in, Signs it is going well. Then give a half-page weekly rhythm for the first two weeks.

Fill in: ROLE_TITLE TEAM_GOALS MANAGER_STYLE TOOLS_AND_SYSTEMS KEY_PEOPLE

Hand to the hiring manager on the new hire's first day and adjust together.

Run in ChatGPT Run in Claude Run in Gemini Prompt page
Example output · Claude (example written by The Prompt Index)
Sample inputs (fictional)
- ROLE_TITLE: Customer Success Manager
- TEAM_GOALS: Reduce average first-response time on support tickets; launch a quarterly business review for the top accounts
- MANAGER_STYLE: Hands-on, but only a few hours a week free; prefers written updates
- TOOLS_AND_SYSTEMS: Zendesk, HubSpot, Notion, Slack
- KEY_PEOPLE: Priya (Support Lead), Tom (Product), Dana (Finance, renewals)

| Phase | Goals | People and meetings | Deliverables | Manager check-in | Signs it is going well |
|---|---|---|---|---|---|
| Days 1-30: learn and listen | Understand the product, the top accounts and the support process | Weekly one-to-one with manager; an introductory conversation with Priya, Tom and Dana in the first fortnight | Shadow support tickets and customer calls [CONFIRM: whether the new hire may speak to customers yet]; write a one-page "what I have learned" note | "Which account worries you most, and why?" | Explains the product in plain words without notes; asks questions in writing instead of staying stuck |
| Days 31-60: contribute with support | Look after a small set of accounts alongside a colleague | Fortnightly check-in with Priya on ticket handling; a session with Dana on renewals [CONFIRM: Dana's availability] | Own follow-ups for the assigned accounts; draft the quarterly review template and test it with one friendly account | "What is slowing you down that I can remove?" | Follow-ups sent without reminders; colleagues use the template draft |
| Days 61-90: own something small | Run a first quarterly review independently | Debrief with manager after the review; feedback from the account | Deliver one review; propose two process improvements with the evidence behind them | "What would you change about how we run these?" | The customer's feedback on the review is constructive; the new hire can say what they would do differently |

Weekly rhythm, weeks 1 and 2
- Week 1: Monday, 30-minute plan for the week with manager. Tuesday to Thursday: system access checks, product walkthrough, shadowing. Wednesday: short written update to manager. Friday: 30-minute reflection ("what surprised you, what is unclear").
- Week 2: Monday plan and review of the "what I have learned" draft. Mid-week: introductory conversations with Priya, Tom and Dana. Friday: reflection, and agree which accounts to shadow in weeks 3 and 4.

Marked [CONFIRM] above: customer contact rules, Dana's availability. Also confirm access to the renewals spreadsheet before day 31.

Remote New Hire First-Week Schedule

Best for: Preventing a lonely remote first week

You are an HR operations lead onboarding a fully remote employee across time zones. Build a first-week schedule that avoids a lonely, tool-heavy start.

New hire's role: {{ROLE_TITLE}}
New hire's time zone and team's core hours: {{TIME_ZONES}}
Tools and accounts needed on day one: {{DAY_ONE_TOOLS}}
Team rituals already in place: {{TEAM_RITUALS}}
Manager's availability this week: {{MANAGER_AVAILABILITY}}

Create a day-by-day schedule for five days with times expressed in the new hire's local time. Include: orientation, equipment and access checks, one-to-one with manager, meetings with two or three colleagues, a buddy session, a first small task, and a Friday reflection.

Constraints: no more than four hours of live meetings on any day; build in focus time and breaks; plan a fallback for the case where equipment or accounts arrive late. Don't assume tools I haven't listed. Note anything that needs to be arranged before the start date.

Output: a table with columns Day, Time (local), Activity, Who, Purpose. Then provide a list of pre-start tasks for HR and IT, and three questions for the Friday reflection.

Fill in: ROLE_TITLE TIME_ZONES DAY_ONE_TOOLS TEAM_RITUALS MANAGER_AVAILABILITY

Send the schedule to the new hire and manager the week before the start date.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

30 and 90 Day New Hire Check-In Questions

Best for: Spotting early-tenure problems while they are fixable

You are an HR business partner designing check-ins with new hires to find problems early, while there is still time to fix them.

Role and team: {{ROLE_AND_TEAM}}
What the job ad or offer promised: {{PROMISED_EXPERIENCE}}
Known onboarding weaknesses: {{KNOWN_WEAKNESSES}}
Who will run the conversation (HR, manager, skip-level): {{INTERVIEWER}}

Create two conversation guides, one for day 30 and one for day 90. Each should have 8 to 10 open questions that cover role clarity, workload, tools, manager relationship, belonging, and whether the job matches what was described.

For each question add a one-line note on what a worrying answer sounds like and a suggested follow-up. Add a short opening script that explains what will happen with the answers, and a closing script that confirms next steps.

Constraints: don't ask leading questions; don't promise confidentiality we can't keep; tell the person that serious concerns such as harassment must be taken further. Remind me to record only what is needed and store notes securely.

Output: two guides in numbered lists, then a one-page template for recording themes and actions that has no personal data beyond the employee's initials.

Fill in: ROLE_AND_TEAM PROMISED_EXPERIENCE KNOWN_WEAKNESSES INTERVIEWER

Run these as conversations, not forms; the manager should not be the only listener.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Offboarding Checklist by Role and Access

Best for: Making sure access and handover are not forgotten

You are an HR operations specialist building an offboarding checklist for a departing employee, so nothing is missed on the last day.

Role and level: {{ROLE_AND_LEVEL}}
Reason for leaving (resignation, end of contract, redundancy, dismissal): {{DEPARTURE_TYPE}}
Systems, accounts and physical assets they hold: {{ACCESS_AND_ASSETS}}
Last working day: {{LAST_DAY}}
Country or state: {{COUNTRY_OR_STATE}}

Produce a checklist grouped by owner (HR, IT, Finance, Manager, Facilities, Employee). For each item add the timing (before the last day, on the last day, after) and what evidence of completion to record.

Include: final pay and holiday balance review, benefits and pension notices, return of equipment, access removal (do this on a timed basis that respects the departure type), handover of work, customer and supplier notifications, references policy, exit interview, reminder of continuing obligations in the contract, and records retention.

Constraints: the correct steps for final pay, notice, benefits continuation and records vary by location and departure type. Tag those [VERIFY LOCALLY] rather than stating rules from memory. Keep tone neutral.

Output: a table with columns Task, Owner, Timing, Evidence, Notes. Add a short list of what to do differently for a dismissal or redundancy.

Fill in: ROLE_AND_LEVEL DEPARTURE_TYPE ACCESS_AND_ASSETS LAST_DAY COUNTRY_OR_STATE

Adapt the timing for involuntary exits and have legal review final-pay items.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Knowledge Handover Plan for a Departing Employee

Best for: Capturing know-how before someone leaves

You are a team operations coach helping a manager capture what a departing employee knows before they go.

Departing person's role: {{ROLE}}
Days remaining: {{DAYS_REMAINING}}
Key responsibilities and recurring tasks: {{RESPONSIBILITIES}}
Who is taking over what: {{SUCCESSORS}}
Critical relationships (clients, suppliers, partners): {{KEY_RELATIONSHIPS}}

Create a handover plan that fits the remaining time. Include:
1. A prioritised list of knowledge to capture, split into must-capture, should-capture and nice-to-have.
2. A suggested format for each (checklist, short recorded walkthrough, shared document, shadowing session).
3. A day-by-day or week-by-week schedule that leaves the departing person room to do their normal work.
4. Questions to ask about undocumented workarounds, passwords stored the wrong way, promises made to clients and risks nobody has written down.
5. A relationship-handover script for introducing successors to key contacts.

Constraints: assume goodwill but plan for the case where the person is distracted or reluctant. Do not ask them to share anything outside their contract duties. Remind the manager to remove access only after handover is complete and to check what the departing person is allowed to share.

Output: a prioritised table and a one-page schedule.

Fill in: ROLE DAYS_REMAINING RESPONSIBILITIES SUCCESSORS KEY_RELATIONSHIPS

Start the same day notice is given; handover quality drops quickly in the last week.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Exit Interview Guide and Theme Summary

Best for: Turning exit interviews into usable themes

You are an HR business partner who runs exit interviews and turns them into evidence for retention decisions.

Part 1 — Interview guide. Context: {{COMPANY_PROFILE}}. Create 10 neutral, open questions for a departing employee on role, manager, growth, pay and recognition, workload, culture, and what would have kept them. Add an opening statement explaining how the answers will be used, and a prompt for when someone says they'd rather not discuss something.

Part 2 — Theme summary. Here are anonymised notes from recent exits (no names, one entry per leaver): {{EXIT_NOTES}}

Group the notes into themes. For each theme state how many entries mention it, quote a short anonymised example, and separate what the leaver said from your interpretation. Do not assume a cause when the evidence is thin, and say how many entries you would need before treating it as a pattern. Don't compute percentages from a very small group.

Constraints: if any note describes harassment, discrimination, safety concerns or other wrongdoing, put it at the top as "needs separate follow-up by HR", and do not include identifying detail. If EXIT_NOTES is empty, do Part 1 only. Say how many entries you would want before calling something a pattern (a rule of thumb).

Output: the guide, then a theme table with columns Theme, Count, Example, Evidence strength, Suggested follow-up.

Fill in: COMPANY_PROFILE EXIT_NOTES

Use Part 1 before each interview and Part 2 each quarter.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Probation Review and Confirmation or Extension Letter

Best for: Fair probation reviews with a documented outcome

You are an HR business partner helping a manager run a probation review fairly and document the outcome.

Probation terms pasted from the contract or policy: {{PROBATION_TERMS_TEXT}}
Expectations set at the start: {{EXPECTATIONS}}
Evidence so far (work done, feedback, concerns), with dates: {{EVIDENCE}}
Manager's proposed outcome (confirm, extend, or not confirm): {{PROPOSED_OUTCOME}}
Country or state: {{COUNTRY_OR_STATE}}

Do this:
1. Map the evidence to each expectation and rate how well it supports the proposed outcome (Strong, Partial, Thin). Mark missing evidence [NEED EVIDENCE].
2. Suggest questions for the review meeting.
3. Draft the outcome letter for confirmation or extension, using only facts given. For an extension, state the reasons, what needs to change and what support is offered. For non-confirmation, stop and tell me to take HR or legal advice before drafting anything.
4. List what to check first: whether the contract or policy allows the outcome, whether notice or process applies, and whether the person has protected status, leave or a pending complaint.

Constraints: do not invent probation rules, periods or notice terms; if not given, write [CONFIRM IN CONTRACT]. Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice.

Output: an evidence table, the question list, the letter and the check list.

Fill in: PROBATION_TERMS_TEXT EXPECTATIONS EVIDENCE PROPOSED_OUTCOME COUNTRY_OR_STATE

Run the review before the probation end date and keep notes factual.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Performance & Difficult Conversations

Prompts for managers and the HR people who coach them: reviews built from evidence, a bias check, fair goals, underperformance scripts, improvement plans, calibration and 360 synthesis, plus a role-play partner. The conversation, plan, warning and dismissal-meeting prompts form a sequence for one case, so use them in order and stop at any point where the prompt tells you to involve HR or counsel.

Performance Review From Manager Notes

Best for: Turning rough notes into a fair, evidence-based review

You are an HR business partner helping a manager write a fair, specific performance review from their own notes.

Employee's role and level: {{ROLE_AND_LEVEL}}
Review period: {{REVIEW_PERIOD}}
Manager's raw notes (achievements, issues, feedback received, dates): {{MANAGER_NOTES}}
Rating scale and what each rating means in our company: {{RATING_SCALE}}
Goals set at the start of the period: {{AGREED_GOALS}}

Write the review with these sections: summary; results against each goal, with the evidence from the notes; strengths with examples; development areas with examples; a suggested rating and why, tied to the scale's definitions (a suggestion only; the manager decides); three goals for the next period.

Constraints: use only facts in the notes. If the evidence for a statement is missing, write [NEED EVIDENCE] instead of inventing it. Describe behaviour and outcomes, not personality. Avoid vague phrases like "a good attitude" without an example. Do not mention health, family, age or other protected characteristics; if the notes include any, flag them for removal. Use labels like Employee A and keep personal data (health details, IDs, addresses) out.

Output: the review in section headings, followed by a list of every [NEED EVIDENCE] item and a one-paragraph consistency check against the rating scale.

Fill in: ROLE_AND_LEVEL REVIEW_PERIOD MANAGER_NOTES RATING_SCALE AGREED_GOALS

Manager edits afterwards; the evidence list tells you what to add.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Bias Check on Review Language

Best for: Catching vague or biased wording before delivery

You are an experienced HR reviewer checking performance review text for bias, vagueness and risk before it is shared.

Review text: {{REVIEW_TEXT}}
Role and level of the employee: {{ROLE_AND_LEVEL}}
Company rating scale: {{RATING_SCALE}}

Analyse the review and produce:
1. Vague or unevidenced statements (for example "not a team player") with the sentence quoted and what evidence would make it fair.
2. Language that may reflect bias, for example personality words used for some groups and skill words for others, comments on tone or style, age-coded or gender-coded wording, and references to family or health. Quote it and explain the concern in one sentence. Say clearly that you are flagging possible patterns for human judgment; you cannot determine intent.
3. Mismatches between the narrative and the rating.
4. A rewrite of each flagged sentence that keeps the manager's intended point but focuses on observable behaviour and results.
5. Anything that could create legal exposure if the review were used in a later dispute, with a recommendation to check with HR or counsel.

Constraints: don't add facts. Where the original is fine, say so. Use labels like Employee A and keep personal data (health details, IDs, addresses) out.

Output: a table with columns Quote, Issue, Suggested rewrite, Evidence needed.

Fill in: REVIEW_TEXT ROLE_AND_LEVEL RATING_SCALE

Run on a draft before it's shared. A human still decides what changes.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

SMART Goals and Key Results for a Role

Best for: Setting measurable, fair goals

You are a performance management coach helping a manager set goals that people can actually act on.

Role: {{ROLE_TITLE}}
Team or company priorities for the period: {{PRIORITIES}}
Employee's current strengths and development areas: {{STRENGTHS_AND_GAPS}}
Time period: {{TIME_PERIOD}}
Measures we can already track: {{AVAILABLE_MEASURES}}

Write 4 to 5 goals. For each give: the goal in one sentence; why it matters to the team; how success will be measured using only measures that exist or can be collected cheaply; a mid-point checkpoint; the support or resources needed; and what is explicitly out of scope.

Constraints: at least one goal should be developmental rather than output-based. No goal should rely on factors the employee cannot influence. Avoid invented numerical targets: where a target needs data I haven't given, write [TARGET: agree with employee after checking last period's baseline]. Keep language plain and avoid buzzwords.

Output: a table with columns Goal, Why it matters, Measure, Checkpoint, Support, Out of scope. Finish with three questions the manager should ask the employee when agreeing the goals together, so the goals are not imposed.

Fill in: ROLE_TITLE PRIORITIES STRENGTHS_AND_GAPS TIME_PERIOD AVAILABLE_MEASURES

Agree the goals in conversation. Do not send the table as a surprise.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Script for an Underperformance Conversation

Best for: Preparing for a first performance conversation

You are an HR business partner coaching a manager through a first formal conversation about underperformance. The goal is clarity and a path to improvement, not building a case.

Employee's role: {{ROLE}}
The gap between expected and actual performance, with dates and examples: {{PERFORMANCE_GAP}}
Support and feedback already given: {{PRIOR_SUPPORT}}
Likely reactions: {{LIKELY_REACTIONS}}
Country or state: {{COUNTRY_OR_STATE}}

Write:
1. A 10-minute preparation checklist for the manager.
2. A conversation outline with opening words, how to state the gap factually, how to ask for the employee's view, how to listen for obstacles (workload, unclear expectations, health, personal circumstances), and how to agree next steps and a follow-up date.
3. Sample wording for three difficult moments: the employee gets emotional, disputes the facts, or says they are being targeted.
4. A list of things the manager must not say, such as threats, comparisons with named colleagues, or guesses about reasons.
5. A short follow-up note template that summarises what was agreed.

Constraints: if the employee mentions health, disability, pregnancy, discrimination or a complaint, the manager should pause and contact HR. Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice. Use labels like Employee A and keep personal data (health details, IDs, addresses) out.

Output: the five items in order, in plain language.

Fill in: ROLE PERFORMANCE_GAP PRIOR_SUPPORT LIKELY_REACTIONS COUNTRY_OR_STATE

Manager rehearses with HR before the meeting; keep the follow-up note factual.

Run in ChatGPT Run in Claude Run in Gemini Prompt page
Example output · Claude (example written by The Prompt Index)
Sample inputs (fictional)
- ROLE: Accounts Payable Clerk, in post for eighteen months
- PERFORMANCE_GAP: Over two months, 9 of 40 sampled invoices were coded to the wrong cost centre (for example the 3 September batch); month-end close was late twice
- PRIOR_SUPPORT: Informal reminder in August; checklist shared; no formal conversation yet
- LIKELY_REACTIONS: Defensive; may say the workload is too high
- COUNTRY_OR_STATE: England (UK)

1. Preparation checklist (10 minutes)
- Bring the dated examples, the checklist and the team's workload for the period.
- Decide the outcome you want: agreed expectations and a support step, not a verdict.
- Check with HR first if Employee A has a pending complaint, health issue or leave.

2. Conversation outline
- Opening: "I would like to talk about invoice coding and the month-end deadline. I want us to agree how to fix it."
- State the gap: "In the sample I checked, 9 of 40 invoices had the wrong cost centre, for example the 3 September batch."
- Ask: "What is your view? What gets in the way?"
- Listen for: unclear rules, workload, system problems, health or personal circumstances.
- Agree next steps: one clear standard, one support step (for example a walkthrough of the cost-centre rules), and a follow-up date.

3. Three difficult moments
- Emotional: "Let us pause. Would you like a few minutes?"
- Disputes the facts: "Let us look at the examples together. If one is wrong, I will correct it."
- Says they are targeted: "I am raising this because of the invoices. If you feel it is more than that, tell me and I will involve HR."

4. Do not say
"You will be out if this continues"; "Everyone else gets this right"; guesses about why; comparisons with named colleagues.

5. Follow-up note
"Thank you for talking today. We agreed [standard]. I will [support]. We will meet on [date]. If anything changes before then, tell me."

Legal flags for England: if Employee A mentions health, disability, pregnancy or discrimination, pause and involve HR. Verify with an employment lawyer or your own capability procedure whether this conversation counts as a formal stage, and what guidance such as Acas applies. This is not legal advice.

Performance Improvement Plan Draft

Best for: Drafting a PIP that is specific and defensible

You are an HR business partner drafting a fair, specific performance improvement plan (PIP) that genuinely aims to help the person succeed. HR and, where needed, counsel will review it.

Role: {{ROLE}}
Specific performance issues with dated examples: {{ISSUES_WITH_EXAMPLES}}
Previous coaching or warnings: {{PRIOR_STEPS}}
Review period length: {{PLAN_LENGTH}}
Country or state: {{COUNTRY_OR_STATE}}

Draft a PIP with: a neutral opening; each expectation stated as an observable standard; the evidence of the current gap; what success looks like at the end; support the company will provide; scheduled check-ins with dates; how progress will be recorded; what happens at the end, without promising an outcome.

Constraints: do not include anything not supported by the examples. If the employee has raised a complaint, has a health condition, or is on or recently back from protected leave, stop and tell me to consult HR or counsel first. Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice. Use labels like Employee A and keep personal data (health details, IDs, addresses) out.

Output: the draft plan, a pre-issue checklist for HR, and a list of questions counsel should review.

Fill in: ROLE ISSUES_WITH_EXAMPLES PRIOR_STEPS PLAN_LENGTH COUNTRY_OR_STATE

Do not issue before HR and, where appropriate, legal have reviewed it.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Calibration Meeting Evidence Pack

Best for: Entering calibration with evidence rather than impressions

You are an HR analyst preparing a manager for a performance calibration meeting, where ratings across teams are compared for consistency.

Team and period: {{TEAM_AND_PERIOD}}
Rating scale definitions: {{RATING_DEFINITIONS}}
For each employee (use initials only), the proposed rating and the evidence supporting it: {{RATING_EVIDENCE}}

For each person produce a one-paragraph summary of the evidence in the manager's own facts, a note on how strongly the evidence supports the proposed rating against the scale definitions, and one question other managers might reasonably ask.

Then compare the team as a whole: where the evidence for two people looks similar but the ratings differ, where a rating relies on recent events only, where the evidence is mostly effort rather than results, and where a rating depends on a single source.

Constraints: you are checking consistency, not setting ratings; the manager decides. Don't use or infer age, gender, ethnicity, health, family status, or other protected characteristics, and point out if the notes refer to any. Say plainly when the evidence provided is too thin to judge.

Output: a table with columns Initials, Proposed rating, Evidence summary, Support strength (Strong/Partial/Thin), Likely challenge. Add a short list of consistency risks.

Fill in: TEAM_AND_PERIOD RATING_DEFINITIONS RATING_EVIDENCE

Use it to rehearse your case, not to override your judgment.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

360 Feedback Synthesis

Best for: Making 360 feedback usable while protecting anonymity

You are an HR specialist turning raw 360 feedback into a useful, kind and honest summary for one employee.

Employee's role: {{ROLE}}
Competencies or values being assessed: {{COMPETENCIES}}
Feedback from manager, peers, direct reports and others, anonymised: {{FEEDBACK_ITEMS}}
Number of respondents in each group: {{RESPONDENT_COUNTS}}

Do this:
1. Identify themes that appear across at least two respondents, and separate them from single comments.
2. Summarise strengths, then development areas, with an anonymised quote for each theme. Remove anything that could identify the respondent, such as a unique project, phrase or incident.
3. Note where groups disagree and offer two possible explanations for the gap, flagged as hypotheses.
4. Point out comments that are personal, unfair, or about protected characteristics, and recommend leaving them out.
5. Suggest three development conversations the employee could have, and two questions for a coach or manager to open the session.

Constraints: if a group has fewer than three respondents (a rule of thumb, not a legal requirement), merge it with another group or leave it out so no one can be identified. Never reconstruct who said what. Don't present small numbers as statistics.

Output: a structured report with headings and a table of themes with the groups that raised them.

Fill in: ROLE COMPETENCIES FEEDBACK_ITEMS RESPONDENT_COUNTS

Give the employee the synthesis, never the raw comments.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Manager Role-Play Partner for Hard Conversations

Best for: Rehearsing a conversation before the real one

You are a role-play partner for a manager rehearsing a difficult workplace conversation. You will play the employee. After each manager turn you will respond in character, and at the end you will switch to coach.

Scenario: {{SCENARIO}}
Employee personality and likely reaction: {{EMPLOYEE_PERSONA}}
What the manager wants to achieve: {{MANAGER_GOAL}}
Manager's level of experience: {{EXPERIENCE_LEVEL}}

Rules for the role play:
1. Stay in character, with realistic reactions (defensive, quiet, upset, argumentative). Do not make the employee unrealistically easy or impossibly hostile.
2. Respond in no more than four sentences per turn so the manager can keep going.
3. Do not invent facts about the employee's private life or health; if the manager asks something that is inappropriate or risky, such as asking about a medical condition, reply as the employee might, then flag it later.
4. When I type "STOP", step out of character and give feedback.

Feedback format: three things that worked, three things to change with alternative wording, any statements that created legal or fairness risk, and a rating from 1 to 5 on clarity, empathy and follow-through with reasons. Remind the manager that real conversations should involve HR when sensitive topics come up.

Begin by briefly describing the setting and asking the manager to open the meeting.

Fill in: SCENARIO EMPLOYEE_PERSONA MANAGER_GOAL EXPERIENCE_LEVEL

Practise three times before a real meeting, changing the persona each time.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Promotion or Pay Review Case Write-Up

Best for: Writing a case that decision-makers can compare fairly

You are an HR business partner helping a manager write a case for a promotion or pay review, in a format that decision-makers can compare fairly with other cases.

Current role: {{CURRENT_ROLE}}
Proposed role: {{PROPOSED_ROLE}}
Evidence of performance at the next level (projects, scope, results, feedback): {{EVIDENCE}}
Role requirements for the next level: {{LEVEL_REQUIREMENTS}}
Time in role and any prior reviews: {{TENURE_AND_HISTORY}}
Budget or band constraints known: {{BUDGET_CONSTRAINTS}}

Write a one-page case with: summary and recommendation; evidence mapped to each requirement of the next level, with strength of evidence (Strong, Partial, Gap); sustained impact rather than one-off wins; risks and how they would be mitigated; and a development plan for any gaps.

Constraints: use only the evidence provided; mark missing evidence [NEED EVIDENCE]. Do not argue from loyalty, tenure alone, or how much the person needs the money. Avoid comparisons with named colleagues. Do not suggest a specific salary figure unless I give you a band; instead list what should be checked against the pay structure and market data. Pay-equity and equal-pay rules vary by place, so tell me to review consistency with peers in similar roles and verify with HR or counsel.

Output: the one-page case, then a short list of the questions a decision-maker is likely to ask.

Fill in: CURRENT_ROLE PROPOSED_ROLE EVIDENCE LEVEL_REQUIREMENTS TENURE_AND_HISTORY BUDGET_CONSTRAINTS

Share with HR before submitting to check consistency with peers.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Employee Communications

Messages employees read closely: reorganisation announcements, benefits enrolment, manager cascade packs, manager cascades, recognition and replies to delicate or flexible-working requests. The prompts use only the facts you supply and refuse to invent promises.

Announcing a Reorganisation to Employees

Best for: Communicating change honestly without overpromising

You are an internal communications lead working with HR on a message about organisational change. People will read it for what it means for their jobs, so be clear and honest.

What is changing and why: {{CHANGE_SUMMARY}}
Who is affected and how: {{AFFECTED_GROUPS}}
What has been decided and what has not: {{DECIDED_VS_OPEN}}
Dates and next steps: {{TIMELINE}}
Country or state: {{COUNTRY_OR_STATE}}

Write: a leadership announcement of no more than 350 words; a list of five anticipated questions with honest answers using only the facts above; and a note of what must NOT go in writing yet.

Constraints: be direct about what is decided; never imply a decision is final if it is not. No euphemisms like "rightsizing". No promises about job security, outcomes or timelines beyond what I gave you. Where the change could involve redundancy, consultation, notice, works-council or collective-notification duties in {{COUNTRY_OR_STATE}}, stop and tell me to speak to an employment lawyer before anything is sent, and list the questions to ask. Do not name individuals.

Output: the announcement, the Q and A, a short sequence for who hears what and when (leaders, affected staff, everyone else), and a list of legal checks required before sending.

Fill in: CHANGE_SUMMARY AFFECTED_GROUPS DECIDED_VS_OPEN TIMELINE COUNTRY_OR_STATE

Have legal approve it. Tell affected people before the general announcement.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Benefits Enrolment or Renewal Window Emails

Best for: Clear benefits communication during enrolment

You are an HR benefits communicator writing a short email series that helps employees make decisions during a benefits enrolment or renewal window, without sounding like a legal notice.

Enrolment window: {{ENROLMENT_WINDOW}}
What's changing this year: {{BENEFIT_CHANGES}}
Benefits on offer: {{BENEFITS_LIST}}
Where to enrol and who to ask: {{HOW_TO_ENROL}}
Typical employee questions: {{COMMON_QUESTIONS}}

Write four emails: (1) announcement, (2) a plain-English explainer of what changed, (3) a mid-window reminder with a "most common mistakes" list based only on the questions I gave you, (4) a final-day reminder.

Constraints: each email under 150 words with one clear call to action and a subject line under 50 characters. Use only details I provided; mark gaps [CONFIRM]. Do not give tax, insurance, medical or financial advice or recommend a plan for any individual; direct people to the provider or HR for personal questions. Do not make claims about savings or value. State deadlines exactly as I gave them. Make the writing accessible for people who read English as a second language.

Output: the four emails with subject lines, a one-paragraph text message version of the final reminder, and a list of facts that must be checked before sending.

Fill in: ENROLMENT_WINDOW BENEFIT_CHANGES BENEFITS_LIST HOW_TO_ENROL COMMON_QUESTIONS

Check dates and plan details with your broker or provider before scheduling.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Manager Cascade Pack for a Difficult Message

Best for: Equipping managers to deliver a message consistently

You are an HR communications partner preparing managers to deliver a message to their teams after the all-staff announcement. Managers are nervous and employees will ask questions.

The announcement: {{ANNOUNCEMENT_TEXT}}
Managers' seniority and training level: {{MANAGER_LEVEL}}
Facts that are settled and facts that are not: {{SETTLED_VS_OPEN}}
Team concerns we already expect: {{EXPECTED_CONCERNS}}

Create a manager pack containing:
1. A 5-minute talking script covering what happened, why, what it means for the team, and what happens next.
2. The top 10 questions and recommended answers, using only the facts above; where the answer is unknown, the line "I don't know yet; I'll find out and come back to you by [DATE]".
3. A list of phrases to avoid, such as speculation, blame, or personal opinions about the decision.
4. How to handle an upset employee, a rumour, or a question about someone else's situation.
5. An escalation guide: what to take to HR, and what to record.

Constraints: the manager must not promise anything not in the announcement. Remind them that questions touching on contracts, pay, redundancy or legal rights go to HR. Tone: calm, respectful, plain.

Output: the pack as a short document with headings, followed by a one-page quick-reference card.

Fill in: ANNOUNCEMENT_TEXT MANAGER_LEVEL SETTLED_VS_OPEN EXPECTED_CONCERNS

Brief managers one day before the all-staff message.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Reply to a Sensitive Employee Email

Best for: Replying carefully to a sensitive message

You are an experienced HR advisor drafting a reply to an employee's email. The reply must be kind, accurate and careful not to admit liability, make promises, or pre-judge an outcome.

The employee's email (remove names first): {{EMPLOYEE_EMAIL}}
Facts we can confirm: {{CONFIRMED_FACTS}}
What we can and cannot say yet: {{SAYABLE_LIMITS}}
Relevant policy or process: {{POLICY_OR_PROCESS}}
Country or state: {{COUNTRY_OR_STATE}}

Do this:
1. Identify the type of email (question, complaint, grievance, request for accommodation, whistleblowing, resignation threat, expression of distress) and any urgency. If it describes harassment, discrimination, safety danger, self-harm or a legal claim, say so first, and recommend escalation instead of drafting an ordinary reply.
2. Write a draft reply under 250 words that acknowledges the concern, states what happens next and by when, and uses only the confirmed facts.
3. List the sentences I must not include, such as opinions about people, promises, or admissions.
4. Give a shorter holding reply for use if I need more time.

Constraints: no invented policy details. Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice. Use labels like Employee A and keep personal data (health details, IDs, addresses) out.

Output: classification and urgency, the draft, the holding reply, and a note of what to record in the file.

Fill in: EMPLOYEE_EMAIL CONFIRMED_FACTS SAYABLE_LIMITS POLICY_OR_PROCESS COUNTRY_OR_STATE

Have a second HR person read anything that may become a formal complaint.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Reply to a Flexible Working or Leave Request

Best for: Replying fairly to a request for flexibility

You are an HR advisor drafting a manager's reply to an employee's request for flexible working, a schedule change or unpaid leave. The reply must be fair and consistent, and must not decide legal entitlement.

Request, summarised in neutral wording: {{REQUEST_SUMMARY}}
How the work is covered today: {{COVER_ARRANGEMENTS}}
Relevant policy and business considerations: {{POLICY_AND_CONSTRAINTS}}
Decision made by the manager and HR: {{DECISION}}
Country or state: {{COUNTRY_OR_STATE}}

First check: if the request might relate to disability, pregnancy, caring duties, religion or a statutory right, say so and tell me to involve HR before replying. Then draft:
1. A reply under 200 words that thanks the employee, states the decision and reasons in terms of the work, and names any next step our policy provides, such as a trial period or an appeal route.
2. If the answer is no or partly no, a version that explains which business reasons apply and offers any alternative I listed.
3. A list of facts to confirm before sending, including any response deadline set by local rules or our policy.

Constraints: use only the facts given; mark gaps [CONFIRM]. Do not cite statutes or timeframes unless certain, and tag any you mention [VERIFY]. Not legal advice.

Output: classification, the reply, the alternative reply, and the confirmation list.

Fill in: REQUEST_SUMMARY COVER_ARRANGEMENTS POLICY_AND_CONSTRAINTS DECISION COUNTRY_OR_STATE

Send only after HR has checked consistency with similar past requests.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Engagement, Surveys, Fairness & Retention

Run a survey people trust, analyse the comments without exposing the authors, turn results into a short action plan, and learn why people stay. A DEI data check works on aggregated tables only. The retention prompts work at team level and deliberately avoid building lists of individuals who might leave.

Pulse Survey Question Builder

Best for: A short survey that employees trust

You are a people-analytics specialist designing a short pulse survey that employees will actually complete and trust.

Topic to learn about: {{SURVEY_TOPIC}}
Population and size (smallest team size reported on): {{POPULATION}}
Decisions the results will inform: {{DECISIONS}}
How the survey will be run (tool, timing, anonymity): {{SURVEY_METHOD}}

Write a survey of no more than 10 items using a 5-point agreement scale, plus two open-text questions. Avoid leading, double-barrelled or loaded questions. For each item state what it measures and what you'd do with a low score.

Also provide: an introduction text that is honest about anonymity and what happens next; a rule for the minimum group size before results are shown, so no team can be identified (recommend a threshold and explain the reasoning without asserting it as a legal requirement); a communication plan for before and after; and a note on what not to ask, such as questions that invite disclosure of health, union membership or other protected details.

Constraints: do not promise anonymity if the tool allows individual identification. Remind me to check data-protection and works-council requirements for surveys where I operate, and to avoid collecting data I don't need.

Output: the survey as a numbered list, the intro text, and a one-page rollout plan.

Fill in: SURVEY_TOPIC POPULATION DECISIONS SURVEY_METHOD

Always commit to sharing results and actions, otherwise the next survey gets lower response.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Open-Comment Theme Analysis With Privacy Guardrails

Best for: Summarising survey comments without exposing authors

You are a people-analytics specialist summarising free-text survey comments for leadership while protecting the people who wrote them.

Survey topic and population: {{SURVEY_CONTEXT}}
Total responses and comment count: {{RESPONSE_COUNTS}}
Comments (strip names, teams, dates and any identifying details first): {{COMMENTS}}

Do this:
1. Group comments into themes. For each theme give a name, a one-sentence description, the number of comments mentioning it, and two short illustrative quotes edited to remove anything identifying.
2. Separate strongly felt, frequent themes from minority views that may still matter (for example safety or fairness issues).
3. Flag any comment that mentions harassment, discrimination, safety risks, self-harm or wrongdoing, without quoting it, and recommend that HR follows up through the proper channel.
4. Note contradictions between themes and what additional information would resolve them.
5. Suggest three questions for leadership to discuss, framed as questions, not conclusions.

Constraints: count only what is there, state clearly that counts from open text are approximate, and do not report percentages for groups with fewer than 10 comments (a rule of thumb). Don't speculate about who wrote a comment or infer demographic details.

Output: a theme table with columns Theme, Approx. count, Example quote, Concern level, Suggested next step, then a half-page executive summary.

Fill in: SURVEY_CONTEXT RESPONSE_COUNTS COMMENTS

Anonymise the comments before pasting and check the summary against the raw text.

Run in ChatGPT Run in Claude Run in Gemini Prompt page
Example output · Claude (example written by The Prompt Index)
Sample inputs (fictional)
- SURVEY_CONTEXT: Annual engagement survey, logistics company; shift workers and office staff
- RESPONSE_COUNTS: 96 responses, 41 comments
- COMMENTS: 41 anonymised comments (names, teams and dates removed)

1. Themes
| Theme | Approx. count | Example quote | Concern level | Suggested next step |
|---|---|---|---|---|
| Late shift-change notice | about 14 | "We hear about changes the night before." | High (most frequent) | Ask operations what the notice practice is and how often it is missed |
| Equipment and tools | about 8 | "The scanners die mid-shift." | Medium | Request a fault log |
| Recognition | about 6 | "Hard to tell if anyone notices." | Medium | Explore in a focus group |
| Training for new starters | about 4 | "I learned by watching." | Medium | Compare with onboarding records |
| Supervisor conduct | 2 | not quoted | Needs HR follow-up | See item 3 |

Counts are approximate. Two comments fit two themes, so the table total is 34 mentions across 32 comments; the other 9 comments were general or praise and are not themed. I have not reported percentages: only the largest theme reaches 10 comments.

2. Frequent versus minority
Frequent and strongly felt: shift-change notice. Minority views that may still matter: supervisor conduct (2 comments) and the new-starter training gap, which is small but consistent.

3. Flagged for HR
Two comments describe a supervisor's language in terms that could indicate a conduct problem. I have not quoted them. HR should read the originals through the proper channel and decide whether a formal process is needed. Two comments do not prove a pattern, and should not be dismissed either.

4. Contradictions
Some comments praise flexibility while others criticise late changes. They may concern different teams; the data cannot show this.

5. Questions for leadership
- What is our stated notice period for shift changes, and who can override it?
- Which areas have the equipment problems?
- What would recognition that people notice look like on the floor?

Executive summary (half page)
The comments point to one main issue: schedule changes that arrive too late for people to plan their lives. It is the largest theme and the one most strongly worded. Equipment reliability comes second and sounds fixable once we know which sites are affected. Recognition and new-starter training are smaller but consistent. Two comments raise a supervisor conduct concern; they appear in the table by count only, are not quoted, and need separate, confidential handling by HR. These counts come from open text and are approximate, and nothing here says which teams or people are involved. Before acting, check the picture against operational data on shift changes and equipment faults, and tell staff what we heard and what we will do.

Survey Results to Action Plan

Best for: Closing the loop on survey feedback

You are an HR business partner helping a leadership team turn engagement results into a small number of actions that are actually delivered.

Headline results (scores, themes, response rate): {{RESULTS_SUMMARY}}
Last year's actions and what happened to them: {{PREVIOUS_ACTIONS}}
Constraints on budget, time and manager capacity: {{CONSTRAINTS}}
Leadership priorities: {{LEADERSHIP_PRIORITIES}}

Do this:
1. Choose no more than three focus areas, explaining why those rather than the others, based only on the results provided.
2. For each, write a root-cause hypothesis and one question to test with employees (for example in a focus group) before acting.
3. Suggest two or three actions per focus area, rated by effort and visibility, including at least one that can be done within 30 days.
4. Define a simple success indicator per action that does not just repeat the next survey score.
5. Propose a communication plan: what we heard, what we will do, what we will not do and why, and a date for the update.

Constraints: stay within the constraints. Do not invent benchmarks or industry comparisons. If results point to a legal or safety issue, say it needs separate handling by HR and counsel.

Output: an action-plan table with columns Focus area, Hypothesis, Action, Owner, Effort, Date, Indicator, then a draft "You said, we did" message.

Fill in: RESULTS_SUMMARY PREVIOUS_ACTIONS CONSTRAINTS LEADERSHIP_PRIORITIES

Share the plan within a month of the results, or trust drops.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Stay Interview Guide for Managers

Best for: Learning why good people stay or leave

You are a talent and retention coach writing a stay interview guide. A stay interview is a one-to-one conversation with a current employee to learn what keeps them, what might push them out, and what to improve.

Manager and team context: {{TEAM_CONTEXT}}
Roles of the employees to be interviewed: {{EMPLOYEE_ROLES}}
Things we can realistically change: {{REALISTIC_CHANGES}}
Things we cannot change: {{OFF_LIMITS}}

Create:
1. A manager briefing explaining the purpose, how to set up the meeting so it does not feel like a warning, and how to listen without defending.
2. Eight to ten open questions (what the person enjoys, what they would change, what they'd miss, what could make them leave, what they want to learn) with a follow-up for each.
3. A closing script that summarises what you heard, commits only to what you can deliver, and agrees when you will follow up.
4. A note for recording: three bullet points only, and what must not be written down (speculation, health, family matters).
5. A way to review patterns across several interviews without naming individuals.

Constraints: the manager must not promise pay rises, promotions or schedule changes during the meeting. No questions about plans to have children, health or age.

Output: the briefing, the question list, scripts, and a one-page recording template.

Fill in: TEAM_CONTEXT EMPLOYEE_ROLES REALISTIC_CHANGES OFF_LIMITS

Run with someone you want to keep, before any resignation signal.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Team Retention Review Without Profiling Individuals

Best for: Understanding retention risk responsibly

You are an HR business partner helping a department head look at retention risk at team level, using aggregated information only. We are not building a list of employees likely to leave.

Team size and structure: {{TEAM_STRUCTURE}}
Turnover in the past 12 months by role (counts only): {{TURNOVER_BY_ROLE}}
Engagement or pulse results for the team: {{ENGAGEMENT_RESULTS}}
Known changes ahead (reorganisation, pay review, leadership change): {{UPCOMING_CHANGES}}
Workload and overtime patterns: {{WORKLOAD_PATTERNS}}

Do this:
1. Describe the retention picture plainly; say when a sample is too small to show a pattern.
2. List the structural risk factors (workload, pay position, career paths, manager span of control, uncertainty) supported by the data given, and the ones that are speculation.
3. Suggest team-level actions: what to ask people, what to fix, what to communicate.
4. List the data we do not have that would change the conclusion.
5. State clearly what we should not do, including predicting which individuals will leave, using private data or messages, or treating people differently because we think they may resign.

Constraints: no inferences about age, family plans, health or other protected characteristics. Remind me to check data-protection rules and be transparent with staff about what we analyse.

Output: a one-page brief with headings and a table of risk factors with evidence strength.

Fill in: TEAM_STRUCTURE TURNOVER_BY_ROLE ENGAGEMENT_RESULTS UPCOMING_CHANGES WORKLOAD_PATTERNS

Keep to team-level data. Never turn it into a flight-risk list.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Low-Cost Recognition Programme Design

Best for: Building recognition that is fair and affordable

You are an HR programme designer helping a company build recognition that employees value without a big budget.

Company size and work patterns (office, shift, remote): {{COMPANY_PROFILE}}
Budget per month: {{MONTHLY_BUDGET}}
Values or behaviours to reinforce: {{VALUES}}
Current recognition and what is not working: {{CURRENT_STATE}}

Design a recognition programme with: three types of recognition (peer-to-peer, manager-to-employee, milestone) and what each looks like in practice; criteria that tie recognition to behaviours, not popularity; a process that keeps it fair across teams, shifts and locations; low-cost ideas that fit the budget; a plan to include remote and part-time staff; and a way to review who is being recognised, so we can spot patterns of unequal recognition.

Constraints: do not suggest cash, gifts or equity unless the tax and payroll treatment is checked; where benefits in kind may be taxable in the location, tell me to confirm with payroll or an accountant. No gamification that pits colleagues against each other. Don't claim recognition will raise performance or retention; frame it as a hypothesis to test.

Output: a one-page programme outline, a table of recognition types with cost and effort, a launch plan for the first 60 days, and a short list of ways the programme could go wrong.

Fill in: COMPANY_PROFILE MONTHLY_BUDGET VALUES CURRENT_STATE

Check the tax treatment of any reward with payroll first.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

DEI Data Review and Inclusive Practice Check

Best for: Checking a process for fairness without profiling people

You are a people analyst and inclusion adviser helping an HR team check its own processes for fairness, with care about what data may be collected and used.

Aggregated data we legally and ethically hold (tables only): {{AGGREGATED_DATA}}
Process under review (for example promotion, pay review or performance ratings): {{PROCESS_UNDER_REVIEW}}
Group sizes: {{GROUP_SIZES}}
Country or state: {{COUNTRY_OR_STATE}}

Do this:
1. Say which comparisons the data can support and which groups are too small to report (a rule of thumb: under ten people). Never infer any individual's characteristics.
2. Describe differences in outcomes between groups in plain terms, with no causal claims; list explanations to test, such as role mix, tenure or manager.
3. Review the process itself: criteria, who decides, how consistent the evidence is.
4. Suggest inclusion improvements to the policy or practice, each with an owner and a way to check.
5. Warn about data-protection rules on demographic data and whether pay-gap reporting or positive-action rules apply in {{COUNTRY_OR_STATE}}; mark these [VERIFY LOCALLY].

Constraints: no invented benchmarks; no quotas or decisions about individuals. Not legal advice.

Output: a comparison table with a confidence column, the process review, the actions list and the verify list.

Fill in: AGGREGATED_DATA PROCESS_UNDER_REVIEW GROUP_SIZES COUNTRY_OR_STATE

Use aggregated data only, and decide with legal advice what you may collect.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Training & Development

Find out whether a gap is really a training problem, then build the fix: first-manager programme, micro-lessons from SOPs, policy quizzes, development plans and an evidence-based succession review. No invented course names or effectiveness statistics.

Training Needs Analysis From Role Gaps

Best for: Choosing training that solves the right problem

You are a learning and development partner carrying out a training needs analysis for one team.

Team and purpose: {{TEAM_AND_PURPOSE}}
Skills the work requires now and in the next 12 months: {{REQUIRED_SKILLS}}
Evidence of gaps (errors, feedback, customer complaints, review themes), without names: {{GAP_EVIDENCE}}
Budget and time available for learning: {{BUDGET_AND_TIME}}

Do this:
1. Turn the skills list into a matrix of skills against proficiency levels (Aware, Working, Strong, Can teach).
2. From the evidence, identify which gaps are skills problems and which are caused by something else, such as unclear process, tools, workload or management. Explain your reasoning and say when the evidence is not enough to tell.
3. Rank training needs by business impact and ease of fixing.
4. For each of the top five, propose a learning approach (on-the-job coaching, short course, shadowing, job aid, community of practice) that fits the budget and time.
5. Define how we would know the training worked, using observable changes in the work.

Constraints: do not recommend a vendor or course by name. Don't invent statistics on training effectiveness. Where a training topic is legally required, such as safety or data protection, tell me to confirm frequency and content requirements locally.

Output: a skills matrix, a prioritised needs table, and a 90-day learning plan.

Fill in: TEAM_AND_PURPOSE REQUIRED_SKILLS GAP_EVIDENCE BUDGET_AND_TIME

Check whether a gap is a skills problem before booking training.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

First-Time Manager Training Outline

Best for: Building a first-manager programme from scratch

You are a leadership development designer building a four-session programme for newly promoted managers at a company of {{COMPANY_SIZE}} people.

Audience background: {{AUDIENCE_BACKGROUND}}
Biggest challenges new managers report here: {{MANAGER_CHALLENGES}}
Session length and format: {{SESSION_FORMAT}}
Company policies and tools managers must know: {{POLICIES_AND_TOOLS}}

For each of the four sessions give: the learning objectives; a timed agenda; one practical exercise using a realistic scenario; the common mistakes to bring up; a between-session task to try at work; and a short reflection question.

Topic coverage across the four sessions should include: moving from doing to leading; one-to-ones and feedback; setting expectations and handling underperformance; managers' legal and policy duties (recognising harassment complaints, accommodation requests, leave requests and when to bring in HR).

Constraints: base examples on situations, not made-up statistics. Don't promise results. For legal content, keep to "recognise and escalate" rather than telling managers what the law requires, and remind me to have HR or counsel review it for my location. Keep the language plain.

Output: a programme overview table, a detailed outline for each session, and a facilitator checklist.

Fill in: COMPANY_SIZE AUDIENCE_BACKGROUND MANAGER_CHALLENGES SESSION_FORMAT POLICIES_AND_TOOLS

Pair with a manager buddy or coaching group so skills stick.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Micro-Learning Lesson From an SOP

Best for: Turning a dense procedure into short training

You are an instructional designer converting a standard operating procedure into a five-minute lesson that employees will remember.

SOP or procedure text: {{SOP_TEXT}}
Audience and their current knowledge: {{AUDIENCE}}
Where the lesson will be delivered (phone, desk, shift briefing): {{DELIVERY_CONTEXT}}
Errors we see most often when people follow this procedure: {{COMMON_ERRORS}}

Create: a single learning objective written as an observable behaviour; a scenario-based opening of no more than 60 words; the procedure rewritten into no more than seven steps with the reason for the key ones; a "what goes wrong" section based on the common errors; a three-question check with answers and short explanations; and a job aid summarised in a few lines.

Constraints: do not change the procedure's meaning or leave out safety-critical or compliance-critical steps. If the SOP is unclear or contradictory, list the questions for the process owner instead of guessing. Use plain words and short sentences. Don't add claims about retention rates or learning science.

Output: the lesson script with headings, the three quiz questions, the job aid, and a list of SOP questions needing owner confirmation.

Fill in: SOP_TEXT AUDIENCE DELIVERY_CONTEXT COMMON_ERRORS

Have the process owner check the finished lesson before launch.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Policy Acknowledgement Quiz From a Policy

Best for: Checking understanding, not just acknowledgement

You are a compliance trainer writing a short quiz that checks whether employees understood a policy, rather than whether they clicked "I agree".

Policy text: {{POLICY_TEXT}}
Audience: {{AUDIENCE}}
Pass mark and number of attempts: {{PASS_RULES}}
Real situations where this policy has been misunderstood: {{MISUNDERSTANDINGS}}

Write 8 questions mixing scenario-based multiple choice and true or false. For each: the correct answer, a one-sentence explanation linked to the policy wording, and which section it tests. Include at least three scenarios based on the misunderstandings I listed.

Rules: base every question only on the policy text. If the policy is unclear on a point, do not test it; list it separately as a clarification request. Avoid trick questions and negative double wording. Make distractor options plausible but clearly wrong when the policy is read properly. Vary where correct answers appear.

Also provide: a short feedback message for a wrong answer; a note on accessibility (reading level, screen readers, translation); and a reminder that the quiz confirms understanding of the policy but does not by itself prove legal compliance, so records of results should be kept according to local retention rules.

Output: the quiz, the answer key with explanations, and the clarification list.

Fill in: POLICY_TEXT AUDIENCE PASS_RULES MISUNDERSTANDINGS

Test on three employees before rolling out.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Individual Development Plan Conversation

Best for: Development plans that employees actually own

You are a career development coach helping a manager and employee build an individual development plan (IDP) together.

Employee's role and tenure: {{ROLE_AND_TENURE}}
Employee's stated interests and aspirations: {{ASPIRATIONS}}
Strengths and gaps from recent feedback: {{FEEDBACK_SUMMARY}}
Opportunities that exist in the company (projects, rotations, courses, mentors): {{OPPORTUNITIES}}
Time and budget available: {{TIME_AND_BUDGET}}

Create:
1. A list of 8 questions the manager can ask to understand what the employee wants from the next two years, including what they do not want.
2. An IDP template with three development goals, each with the target skill, an action using the 70-20-10 idea of learning mostly on the job, with others and through courses (explain it simply and say it is a rule of thumb, not a proven formula), a timeframe, the support needed and how progress will be reviewed.
3. A realistic picture of how this plan connects to current work and to possible future roles, without promising promotion.
4. A review schedule.

Constraints: do not make promises about promotion, pay or role changes. Do not assume career goals from age, gender, or caring duties. Keep to the opportunities listed.

Output: the question list, the filled-in draft IDP, and a one-paragraph note to the manager on following through.

Fill in: ROLE_AND_TENURE ASPIRATIONS FEEDBACK_SUMMARY OPPORTUNITIES TIME_AND_BUDGET

Employee drafts first, then the manager adds support.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Succession and Talent Review Preparation

Best for: Preparing a succession discussion that rests on evidence

You are a talent partner preparing a leadership team for a succession review of critical roles. The aim is a discussion based on evidence, not a list of favourites.

Critical roles and why they are critical: {{CRITICAL_ROLES}}
Possible successors by initials, with evidence of performance and readiness: {{SUCCESSOR_EVIDENCE}}
Skills and experience each role requires: {{ROLE_REQUIREMENTS}}
Time horizon: {{TIME_HORIZON}}

Produce:
1. A readiness assessment per successor against the requirements: Ready, Needs development, or Not enough evidence. Flag evidence that is old, single-source, or about potential instead of results.
2. For each role, the risk if the holder left suddenly and an emergency cover idea.
3. A development action per successor, tied to a gap.
4. A bias check on the nominations: who is missing, whether criteria favour people like the nominators, and whether part-time, remote or recently returned staff were considered. Do not infer age, gender, ethnicity, health or family plans.
5. Questions for the leadership discussion.

Constraints: no rankings on potential labels without evidence; ratings are drafts for the leaders to decide. Remind me the list is confidential, nobody should be promised a role, and data-protection rules apply.

Output: a table with columns Role, Successor, Readiness, Evidence strength, Development action; then the bias check and questions.

Fill in: CRITICAL_ROLES SUCCESSOR_EVIDENCE ROLE_REQUIREMENTS TIME_HORIZON

Use for the annual talent review. Do not share ratings with the people named.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

HR Metrics, Reporting & Planning

Define metrics once so everyone calculates them the same way, write turnover and leadership reports from your own numbers, plan headcount scenarios, and draft pay bands from market data you supply. Calculations are shown so you can check them.

HR Metrics Definitions and Dashboard Plan

Best for: Making HR numbers consistent and useful

You are a people-analytics lead helping a small HR team define a basic metrics set so that everyone calculates the same numbers the same way.

Company headcount and structure: {{COMPANY_PROFILE}}
Systems that hold HR data (HRIS, payroll, spreadsheets): {{DATA_SOURCES}}
Questions leadership asks HR most: {{LEADERSHIP_QUESTIONS}}
Reporting frequency: {{REPORTING_FREQUENCY}}

For each of these areas pick the two or three most useful metrics: headcount, turnover (voluntary and involuntary), early-tenure attrition, absence, engagement, learning, and internal mobility.

For each metric write: a plain-English definition; the formula, including what counts as the numerator and denominator; the data source; known pitfalls (for example small numbers swinging a rate, double counting contractors, or unclear leaver dates); which decision the metric supports; and how to display it.

Constraints: do not give industry benchmarks or "good" numbers; instead tell me how to find credible benchmarks and how to compare with ourselves over time. Warn about any metric that could be misused to judge individuals. Remind me to check data-protection rules on what is collected and who can see it.

Output: a table with columns Metric, Definition, Formula, Source, Pitfall, Decision supported, then a simple one-page dashboard layout.

Fill in: COMPANY_PROFILE DATA_SOURCES LEADERSHIP_QUESTIONS REPORTING_FREQUENCY

Agree definitions once and write them down before building any dashboard.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Turnover Analysis Narrative From Aggregates

Best for: Explaining turnover without overclaiming

You are a people analyst writing a short, honest narrative about turnover for leadership, using aggregated numbers only.

Period covered: {{PERIOD}}
Monthly headcount and leavers table: {{MONTHLY_HEADCOUNT_TABLE}}
Leavers by team, tenure band and reason (counts only): {{LEAVER_BREAKDOWN}}
Context: hiring freeze, pay review, reorganisation, seasonal patterns: {{CONTEXT}}

Do this:
1. Calculate the turnover rate and show each division so I can recheck it; models make arithmetic errors. State the denominator you used.
2. Describe patterns by team, tenure and reason, saying where numbers are too small to support a conclusion. Don't draw conclusions from groups of fewer than five leavers (a rule of thumb).
3. Distinguish what the numbers show from what is possible explanation, labelling the second as hypotheses.
4. List the additional data that would improve the analysis.
5. Write a 150-word summary for leadership in neutral wording, followed by three questions leadership should discuss.

Constraints: no invented benchmarks or comparisons with other companies. Do not identify individuals or let small groups be identified. Do not infer protected characteristics. If any figure seems inconsistent, say so rather than adjusting it.

Output: the calculation, a table of breakdowns with a column for confidence, the hypotheses list, and the leadership summary.

Fill in: PERIOD MONTHLY_HEADCOUNT_TABLE LEAVER_BREAKDOWN CONTEXT

Check the arithmetic yourself, as language models can miscalculate.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Headcount Plan and Budget Scenarios

Best for: Planning headcount with clear assumptions

You are an HR finance partner helping a department head think through next year's headcount under different scenarios. You give structure; the finance team confirms the numbers.

Current headcount by role: {{CURRENT_HEADCOUNT}}
Business plan and growth assumptions: {{BUSINESS_ASSUMPTIONS}}
Average fully loaded cost per role, as provided by finance: {{ROLE_COSTS}}
Expected attrition and internal moves: {{ATTRITION_ASSUMPTIONS}}
Hiring constraints (lead time, freezes, approvals): {{HIRING_CONSTRAINTS}}

Build three scenarios (conservative, base, ambitious). For each: headcount by role by quarter; cost by quarter using the figures provided; key assumptions; risks; trigger points that would make us move from one scenario to another; and the effect on workload and management spans.

Constraints: use only the figures I gave; show the arithmetic in a visible table; where a figure is missing, use a labelled placeholder rather than inventing one. Do not recommend redundancies or specific individuals. If a scenario implies reducing roles, flag that notice, consultation and collective-dismissal rules may apply in my location and tell me to involve legal advice before planning.

Output: a scenario comparison table, a list of assumptions to validate with finance and leaders, and a summary of the decision the department head needs to make.

Fill in: CURRENT_HEADCOUNT BUSINESS_ASSUMPTIONS ROLE_COSTS ATTRITION_ASSUMPTIONS HIRING_CONSTRAINTS

Ask finance to check the numbers before sharing.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Monthly People Report for Leadership

Best for: A leadership report that prompts decisions

You are an HR lead writing a concise monthly report for the leadership team that helps them decide, not just read.

Reporting month: {{MONTH}}
Headline numbers (headcount, joiners, leavers, absence, open vacancies): {{HEADLINE_NUMBERS}}
Key events and projects this month: {{KEY_EVENTS}}
Risks or issues HR is monitoring (aggregated, no names): {{RISKS}}
Decisions or support needed from leadership: {{LEADERSHIP_ASKS}}

Write a one-page report with: three-sentence summary; a numbers section with change since last month and a note explaining any change that is more than noise; progress on projects; risks with owner and next step; and a clear "Decisions needed" section.

Constraints: report only the numbers given. Do not add benchmarks or invented trends. Where a movement is based on fewer than five people (a rule of thumb), say it is small. Do not name employees or describe individual cases; use "a case is being handled by HR". Tone: factual, not defensive or promotional. Explain jargon.

Output: the one-page report in markdown, a table version of the numbers, and a list of three questions the leadership team might ask, with short answers using only the information above.

Fill in: MONTH HEADLINE_NUMBERS KEY_EVENTS RISKS LEADERSHIP_ASKS

Keep it to one page; link to detail rather than pasting it.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Pay Band Framework From Market Data You Supply

Best for: Drafting pay bands without inventing market data

You are a reward specialist helping an HR team draft a simple pay band structure. You use only the market data and internal data I provide, and you do not invent salary figures.

Company size and locations: {{COMPANY_PROFILE}}
Job families and levels in scope: {{JOB_FAMILIES}}
Market data points I have gathered (source and date for each): {{MARKET_DATA}}
Current pay ranges and known problems: {{CURRENT_PAY_SITUATION}}
Pay philosophy (for example match market median): {{PAY_PHILOSOPHY}}

Do this:
1. Propose a band structure: number of levels, how wide each band is, and the overlap between levels, with the reasoning.
2. Using only the data I provided, show how each band midpoint would be derived. Mark gaps as [DATA NEEDED].
3. Identify problems such as compression, overlaps that don't make sense, or bands thin on evidence.
4. Describe how to place existing employees in bands and how to handle people paid above or below range, without naming anyone.
5. Explain how to check for unexplained pay differences between groups, and note that equal-pay and pay-transparency rules differ by place.

Constraints: no invented salary numbers or market claims; I verify everything. Tell me to involve an employment lawyer or reward adviser before communicating bands.

Output: a band table with formulae shown, the list of data gaps, and a communication risk list.

Fill in: COMPANY_PROFILE JOB_FAMILIES MARKET_DATA CURRENT_PAY_SITUATION PAY_PHILOSOPHY

Fill the data gaps before using any figures.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Investigations, Accommodations & Employee Relations

The high-risk end of HR. These prompts plan an investigation, organise notes into neutral facts, structure accommodation and absence conversations, draft warning and grievance outcome letters, prepare a dismissal meeting and triage complaints. Each flags legal risk, keeps personal data out and tells you what to verify with an employment lawyer.

Workplace Investigation Plan and Interview Questions

Best for: Planning a fair workplace investigation

You are an experienced HR investigator planning a fair, impartial workplace investigation. Your plan supports a trained investigator and does not replace legal advice or our procedure.

Allegation in neutral wording (no names; use Employee A and Employee B): {{ALLEGATION_SUMMARY}}
Policy or standard the allegation relates to: {{RELEVANT_POLICY}}
Known facts and documents: {{KNOWN_FACTS}}
Country or state: {{COUNTRY_OR_STATE}}

Produce:
1. A scope statement that defines what will and will not be investigated.
2. Immediate steps to consider: interim measures, evidence preservation, confidentiality, protection against retaliation. Don't recommend suspension or discipline; list considerations and who decides.
3. An evidence list: documents, messages, logs and witnesses to gather, and how to record chain of custody.
4. Interview plans for complainant, respondent and a witness, with open, non-leading questions; each person must be able to respond to allegations about them.
5. A timetable, note-taking standard and report structure.
6. Conflict-of-interest and bias checks on the investigator.

Constraints: Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice. Refer to people as Employee A, Employee B. If it involves possible crime, safety danger or senior leaders, tell me to involve counsel first.

Output: a plan with numbered sections and a risk list.

Fill in: ALLEGATION_SUMMARY RELEVANT_POLICY KNOWN_FACTS COUNTRY_OR_STATE

Have counsel or an experienced investigator review the plan before you start.

Run in ChatGPT Run in Claude Run in Gemini Prompt page
Example output · Claude (example written by The Prompt Index)
Sample inputs (fictional)
- ALLEGATION_SUMMARY: Employee A says Employee B repeatedly made demeaning comments about A's accent in team meetings over several months
- RELEVANT_POLICY: Dignity at Work policy, section 3
- KNOWN_FACTS: Two team meeting recordings exist; Employee C and Employee D attended most meetings; no previous complaints on file
- COUNTRY_OR_STATE: California (US)

1. Scope
Whether Employee B's comments in team meetings during the period alleged breached section 3. Out of scope: unrelated performance matters, and conduct outside the period unless new evidence arises.

2. Immediate considerations (for the decision-maker; not recommendations)
- Interim measures short of suspension, such as changes to meeting arrangements.
- Preserve recordings, chat messages and calendar entries.
- Confidentiality instructions that do not stop anyone seeking advice.
- A reminder to everyone involved that retaliation is prohibited. Who decides: [DECISION-MAKER TO BE NAMED].

3. Evidence
Meeting recordings; chat logs; calendar invitations; Employee A's written account; witnesses Employee C and D. Record who collected each item, when and from where.

4. Interview plans
- Employee A: "Tell me in your own words what happened." "When did it start?" "Who was present?" "Is anything written down?" "What outcome would help?"
- Employee B: put each allegation specifically and ask for their account: "What do you recall of the meeting on [DATE]?" Let B respond to every allegation about them.
- Witnesses: "What did you hear and see?" Avoid leading questions such as "Did B insult A?"

5. Timetable, notes and report
Timetable: [SET BY INVESTIGATOR AND DECISION-MAKER, following our procedure and local rules]. Notes: two people present, interviewee checks them for accuracy. Report: allegations, evidence, findings, reasoning.

6. Conflict and bias checks
Investigator is outside both people's reporting lines and has no friendship with either; ask both whether they object to the investigator.

Risk list
- Delay increases memory loss and perceived unfairness.
- Recordings may raise consent and privacy questions [VERIFY].
- Leaks of the complaint could expose the company to retaliation claims.
- The allegation may involve national origin: ask counsel.

Legal flags for California: employer duties on harassment, investigation timing and retaliation are state-specific; verify with California employment counsel. This is not legal advice.

Investigation Notes to Fact Summary

Best for: Organising evidence without prejudging it

You are an HR investigator organising interview notes into a neutral fact summary. You are not deciding whether an allegation is upheld; that decision belongs to the person responsible under our procedure.

Allegation in neutral wording: {{ALLEGATION_SUMMARY}}
Interview notes and documents, with names replaced by labels: {{INTERVIEW_NOTES}}
Policy standard being applied: {{POLICY_STANDARD}}

Do this:
1. Build a timeline of events with the source for each entry (interview, document or message). Mark each entry as Agreed, Disputed or Single-source.
2. List the points where accounts conflict and the evidence supporting each version.
3. List gaps: people not yet interviewed, documents not seen, questions not asked.
4. Separate facts from opinions, assumptions and hearsay.
5. Identify credibility considerations that a decision-maker might weigh, such as consistency, corroboration and motive, expressed as questions and not as conclusions.
6. Draft a neutral structure for the findings section, with headings only.

Constraints: do not reach a finding or recommend a sanction. Do not add facts. Quote notes exactly when something is disputed. Flag legal risk and what to verify locally; this is not legal advice.

Output: a timeline table with columns Date, Event, Source, Status (Agreed/Disputed/Single-source), a conflicts table, a gaps list, and the findings headings.

Fill in: ALLEGATION_SUMMARY INTERVIEW_NOTES POLICY_STANDARD

Check every entry against the original notes. Do not paste in real names.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Accommodation Request Intake and Conversation Notes

Best for: Handling an accommodation request fairly

You are an HR partner supporting a manager who has received a request for a workplace adjustment or accommodation. The goal is a fair, documented, respectful process. Rules on disability, pregnancy and religion differ by location, so you structure the process and do not decide entitlements.

Country or state: {{COUNTRY_OR_STATE}}
Nature of the request, in the employee's own words (no medical details): {{REQUEST_DESCRIPTION}}
Role and key duties: {{ROLE_DUTIES}}
Options the workplace could realistically offer: {{POSSIBLE_OPTIONS}}

Produce:
1. A short script for the first conversation: listen, say thank you, explain the process, say what happens next, and avoid asking about diagnosis.
2. Questions on the practical effect on the work and what has helped, avoiding unnecessary medical information. Say that HR or counsel decides if medical evidence is needed.
3. An options table: adjustment, how it helps, cost, effect on others, what to verify.
4. A response timeline.
5. A record-keeping template that keeps medical information out of the manager's file.

Constraints: Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice. Use labels like Employee A and keep personal data (health details, IDs, addresses) out. Never call a request unreasonable before HR has taken advice.

Output: the five pieces in order.

Fill in: COUNTRY_OR_STATE REQUEST_DESCRIPTION ROLE_DUTIES POSSIBLE_OPTIONS

Escalate to HR immediately. Do not keep medical details in manager files.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Written Warning Letter Draft

Best for: A proportionate, procedure-aware warning letter

You are an HR advisor drafting a written warning that is factual, proportionate and consistent with our disciplinary procedure. HR must check it before issue.

Employee's role (no name): {{ROLE}}
What happened, with dates and facts: {{INCIDENT_FACTS}}
Policy or rule breached: {{POLICY_BREACHED}}
Procedure followed so far (investigation, meeting, employee's response): {{PROCESS_STEPS}}
Warning level and consequence if repeated: {{WARNING_LEVEL}}
Country or state: {{COUNTRY_OR_STATE}}

Draft the letter with: a heading; what the meeting was about; the facts established; the policy breached; the employee's response, summarised fairly; the decision and reasons; the expectations going forward; how long the warning stays on file, using our policy; and any appeal route our procedure provides.

Constraints: use only the facts and procedure I gave; mark any gap [CONFIRM]. No inflammatory language, motive assumptions or references to health. If the process skipped a step such as a hearing, tell me before drafting. Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice.

Output: the letter, then a pre-issue checklist (consistency with similar cases, procedure followed, evidence on file, appeal details) and a short list of legal points to verify.

Fill in: ROLE INCIDENT_FACTS POLICY_BREACHED PROCESS_STEPS WARNING_LEVEL COUNTRY_OR_STATE

Compare against similar past cases to keep treatment consistent.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Dismissal Meeting Preparation and Script

Best for: Preparing a dismissal meeting with the right checks

You are a senior HR business partner helping a manager and HR representative prepare for a meeting where an employee will be dismissed. This is a high-risk task: you prepare the meeting and checklist, and do not decide whether the dismissal is lawful.

Country or state: {{COUNTRY_OR_STATE}}
Reason for dismissal and the procedure completed: {{REASON_AND_PROCESS}}
Employee's service, role and any special circumstances (without medical or personal details): {{EMPLOYEE_CONTEXT}}
Practical arrangements (final pay, benefits, equipment, access): {{PRACTICAL_ARRANGEMENTS}}

Start by listing the checks that must be confirmed with HR or counsel before the meeting happens (for example procedure completed, notice or pay in lieu, protected status, pending complaints or leave, consistency with similar cases, written documentation). If any check is missing, say the meeting should not proceed until it is resolved.

Then provide: who should attend; a short script (under two minutes), clear, respectful and not reopening the decision; responses for three reactions (shock, anger, request for explanation); logistics; what to give the employee in writing; and a follow-up communication plan to the team.

Constraints: Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice. Keep the tone humane.

Output: a pre-meeting checklist, the script, reaction responses, and post-meeting steps.

Fill in: COUNTRY_OR_STATE REASON_AND_PROCESS EMPLOYEE_CONTEXT PRACTICAL_ARRANGEMENTS

Do not hold the meeting until HR and counsel have signed off the checklist.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Employee Complaint Intake and Triage Questions

Best for: Handling the first minutes of a complaint well

You are an HR advisor creating a consistent way to receive and triage employee concerns, so serious issues are not missed.

Company size and existing channels (HR inbox, manager, hotline, anonymous form): {{CHANNELS}}
Policies that apply (grievance, harassment, whistleblowing, safety): {{APPLICABLE_POLICIES}}
Country or state: {{COUNTRY_OR_STATE}}
Who handles what today: {{CURRENT_OWNERSHIP}}

Produce:
1. A triage guide that sorts a concern into: urgent safety or wellbeing risk, possible harassment or discrimination, possible misconduct or wrongdoing, a grievance about treatment, a management or workload issue, a request for support. For each give the first action, who owns it and the target time to respond.
2. An intake script for the first conversation: open questions, what not to ask, how to explain confidentiality honestly (what can and cannot be kept confidential), and how to describe retaliation protection.
3. A one-page intake form that captures facts without opinions or unnecessary personal data.
4. A rule for conflicts of interest and for concerns about HR or senior leaders.
5. A template acknowledging a concern.

Constraints: Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice. Do not decide whether a concern is valid.

Output: the five items in order, using tables where helpful.

Fill in: CHANNELS APPLICABLE_POLICIES COUNTRY_OR_STATE CURRENT_OWNERSHIP

Train every manager on the first two steps, even if HR handles the rest.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Absence Pattern Review and Return-to-Work Conversation

Best for: Handling absence supportively and consistently

You are an HR advisor helping a manager handle sickness absence supportively and consistently. Rules on absence and health information vary by location, so you structure the conversation and do not decide entitlements.

Absence record for one employee (dates and number of days only, no diagnoses): {{ABSENCE_RECORD}}
Our absence policy and any trigger points: {{ABSENCE_POLICY}}
Role and working pattern: {{ROLE_AND_PATTERN}}
Support already offered: {{SUPPORT_OFFERED}}

Produce:
1. A neutral summary of the record against the policy, saying clearly if it does not reach a trigger point, and what the record cannot tell us.
2. A return-to-work conversation guide: welcome back, check they are fit to work, ask what help they need, explain next steps. Questions must not ask for a diagnosis.
3. Signals to stop and involve HR: possible disability or long-term condition, work-related cause, pregnancy, stress or mental health, a pattern linked to a protected characteristic.
4. A short file note limited to dates, the conversation held and agreed adjustments, with medical detail left out.
5. What the manager should not say, such as speculating about the cause.

Constraints: Flag legal risk in {{COUNTRY_OR_STATE}} and what to verify with an employment lawyer; this is not legal advice. Use Employee A for the person, never their name.

Output: the five items in order, in plain language.

Fill in: ABSENCE_RECORD ABSENCE_POLICY ROLE_AND_PATTERN SUPPORT_OFFERED COUNTRY_OR_STATE

Hold the return-to-work conversation promptly after each absence, not only when a trigger is hit.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

Grievance Outcome Letter

Best for: Closing a grievance with a clear, reasoned letter

You are an HR advisor drafting a grievance outcome letter that is clear, fair and does not overstate what was decided. The decision-maker must approve it.

Grievance, in neutral wording: {{GRIEVANCE_SUMMARY}}
Process followed (meetings, evidence considered, who decided): {{PROCESS_FOLLOWED}}
Findings and decision, point by point, as decided by the decision-maker: {{DECISIONS_BY_POINT}}
Actions that will be taken: {{AGREED_ACTIONS}}
Appeal route in our policy, if any: {{APPEAL_ROUTE}}

Draft the letter with: thanks and a summary of the complaint; what was done to look into it; the outcome on each point (upheld, partly upheld or not upheld) with reasons drawn from the evidence given; the actions and owners; the appeal route as our policy describes it; and a reminder about support and protection from retaliation.

Constraints: use only the decisions and reasons given; make no findings yourself. Mark any point lacking reasons [REASONS NEEDED]. Keep the tone respectful, avoid blame, and do not reveal confidential information about others. Flag legal risk for the location and what to verify with an employment lawyer; this is not legal advice. Use Employee A for the complainant.

Output: the letter, then a list of what the decision-maker should confirm before it is sent.

Fill in: GRIEVANCE_SUMMARY PROCESS_FOLLOWED DECISIONS_BY_POINT AGREED_ACTIONS APPEAL_ROUTE

Pair with the Investigation Notes to Fact Summary prompt for the findings.

Run in ChatGPT Run in Claude Run in Gemini Prompt page

How to get the best results

  1. Check your tool first. Before pasting anything about a real employee, confirm your company allows it and that the tool's terms keep data out of model training. When in doubt, anonymise: Employee A, Manager B, no health details, IDs or addresses.
  2. Fill the {{VARIABLES}} with real facts. The more specific the facts, the better the draft. Where you do not know something, say so and let the prompt mark it as a gap.
  3. Read the flags. Items marked [CHECK LOCAL LAW], [VERIFY] or [NEED EVIDENCE] are the model telling you what it cannot know. Resolve each one before using the output.
  4. Keep a human in every decision about a person. Use the output to prepare, draft and organise. Decisions about hiring, pay, discipline and dismissal belong to you, with legal advice where needed.
  5. Test, then reuse. Try a prompt on a past, closed example, adjust the wording to your culture, and save your version.

The worked examples use UK and US scenarios; the legal flags depend on the location you enter. This pack has no job-description, compensation-design or redundancy prompts. Related tools: Helper Studio can turn your finished handbook into a chat helper that answers staff policy questions, and Decision Lab helps structure options before a leadership decision.

Take the whole pack with you

Download all 55 prompts as a tidy PDF, or as Markdown you can paste straight into Notion. It needs a free account: no card, and the prompts above stay free to copy either way.

Already have an account? Log in

Frequently asked questions

Can I use ChatGPT for HR work without breaking confidentiality?

Only if you control what goes in. Use an approved workplace tool whose terms keep your data out of model training, or anonymise first with labels such as Employee A and no health details, ID numbers or addresses. Many companies have an AI policy, so check yours. The Acceptable AI Use Policy prompt in this pack helps you write one.

Are the outputs legal advice?

No. Employment law differs by country and state, and models can be wrong or out of date. The prompts ask the model to flag legal risk and list what to verify. For terminations, investigations, accommodations and anything touching pay equity, have an employment lawyer or qualified adviser review the result before you act.

Can AI decide who to hire, discipline or dismiss?

It should not. Some places restrict automated decision-making about people, and a model can reproduce bias. These prompts help you prepare, draft and organise information, and the decision stays with a person who can explain the reasons.

How is this different from the recruiters pack?

This pack starts when a candidate has accepted an offer and runs through to exit. Sourcing, job adverts, screening and interviewing candidates belong to the recruiters pack.

Will these work for a very small company or a solo HR person?

Yes. The prompts assume limited time and no specialist team. Skip any part that does not fit your size, and use the sensitive-case prompts as preparation before you speak to a lawyer or adviser, not as a substitute.

← All prompt packs