ChatGPT prompts for recruiters: 22 copy and paste prompts that save hours every week

The short version: ChatGPT can cut the time a recruiter spends on job ads, outreach messages, screening questions and client updates by roughly half, if you write prompts that name the role, the seniority, the location and the tone you want. It will not read a CV for you reliably and it should never see a candidate's name, email, phone number or salary history. Below are 22 prompts you can copy, paste and adapt today, plus the mistakes that get recruiters into trouble.

Job ad and job description prompts

A generic job ad prompt gives you a generic job ad. The fix is to feed ChatGPT the same detail you would give a hiring manager on a briefing call: day rate or salary band, the three things that make this role hard to fill, and who reports to whom.

1. "Write a job advert for a [job title] role in [city/region], salary [range], for a [company size] company in [sector]. Highlight [benefit 1] and [benefit 2]. Keep it under 300 words and end with a clear call to action to apply."
2. "Turn this messy set of notes from a client call into a structured job description with sections for responsibilities, requirements, and nice to haves: [paste your notes, no candidate or client personal data]."
3. "Write three different headline options for a LinkedIn job post for a [job title] role, one focused on salary, one on flexibility, one on career growth."
4. "Rewrite this job description to remove jargon and make it sound like a person wrote it, not a template: [paste job description]."
5. "List 10 questions I should ask a hiring manager before writing a job ad for a [job title] role, so I capture what matters most to them."

What to change each time: the job title, the salary band, the sector, and the two benefits you want to lead with. If you skip the sector, ChatGPT will write something bland that could apply to any company.

Candidate outreach and sourcing message prompts

Cold outreach on LinkedIn gets ignored when it reads like a template. These prompts push ChatGPT to write something a candidate would want to reply to, but you still need to add one specific detail about the person you found before you send it.

1. "Write a short LinkedIn message to a passive candidate for a [job title] role. Mention that I found their profile through [source] and that the role offers [one specific benefit]. Keep it under 80 words, no exclamation marks."
2. "Write a follow up message for a candidate who viewed my first LinkedIn message but did not reply. Keep it under 40 words and give them an easy way to say no thanks."
3. "Write a InMail subject line and opening line for a [seniority level] candidate in [sector], aimed at someone who is not actively job hunting."
4. "Write three variations of a boolean search string to find [job title] candidates with [skill] and [years of experience] on LinkedIn."
5. "Draft a short message asking a candidate for a referral to someone else who might suit a [job title] role, to send after a rejection."

Change the source, the sector, and the specific benefit in each message. A candidate can tell within one line whether you have read their profile or copied and pasted, so add the one real detail yourself after ChatGPT gives you the shell.

Screening and interview question prompts

Do not paste a candidate's CV or personal details into ChatGPT to get interview questions. Instead, describe the role and skill you are testing for, and let it build the question bank from that.

1. "Write 10 behavioural interview questions for a [job title] role that test for [skill, e.g. stakeholder management] using the situation, task, action, result format."
2. "Write a screening call script for a [job title] role covering notice period, salary expectations, motivation for moving, and availability, with follow up questions for each."
3. "Build a scorecard template with five criteria for assessing a [job title] candidate at first stage interview, each scored 1 to 5."
4. "Write three follow up questions to ask if a candidate gives a vague answer about why they are leaving their current role."
5. "Turn this list of must have skills into a set of technical screening questions for a [job title] role: [paste skills list, no candidate data]."

Client update and hiring manager email prompts

Client updates are where recruiters lose the most time, rewriting the same shortlist email in a slightly different tone each week. These prompts speed up the first draft. You still add real names and figures yourself once the draft is in front of you, never inside the prompt.

1. "Write a shortlist update email to a client explaining that we have [number] candidates lined up for [job title], with interviews proposed for [week]. Tone: warm but efficient."
2. "Write an email to a hiring manager explaining that the salary they are offering is below market rate for [job title] in [location], with a suggested range based on current demand."
3. "Draft a chasing email to a client who has not given feedback on a candidate in 5 working days, polite but with a gentle deadline."
4. "Write an email letting a client know a strong candidate has another offer on the table and we need a decision within 48 hours."

What to change: the role, the numbers, and the deadline. Read the draft back before sending, since ChatGPT tends to soften urgency, and a chasing email that sounds too polite will not get a reply.

Review reply and reputation prompts

Recruitment agencies live and die on Google and Glassdoor reviews. Replying quickly, in a human tone, matters more than a perfectly worded response.

1. "Write a thank you reply to a five star Google review from a candidate who says we placed them quickly and communicated well throughout."
2. "Write a reply to a two star Glassdoor review from a candidate who says we never gave feedback after their interview. Acknowledge the issue without admitting legal liability."
3. "Write a short reply to a client review praising our shortlist quality, inviting them to refer other hiring managers."

Always read a negative review reply back before it goes live. ChatGPT will sometimes admit fault in language that sounds like an apology for a process failure, and that wording can matter if the review turns into a complaint.

Social post prompts for LinkedIn

1. "Write a LinkedIn post announcing we have 5 open roles in [sector] this month, written from the perspective of a recruitment consultant, not a corporate account."
2. "Write a LinkedIn post sharing one piece of advice for candidates writing a CV for a [sector] role, under 150 words."
3. "Write a LinkedIn post celebrating a candidate placement, without naming the candidate or the client, focused on the outcome."

Admin and reporting prompts

1. "Turn these bullet points about this week's placements and pipeline into a short weekly report for my manager: [paste bullet points, no candidate personal data]."
2. "Write a template email to request feedback from a client after every interview stage, so I can send it without rewriting it each time."

A weak prompt versus a strong prompt

Weak: "Write a job ad for a recruiter." This gives ChatGPT nothing to work with, so it invents a generic company, a generic salary, and generic responsibilities that will not match the real role.

Strong: "Write a job advert for a Senior Recruitment Consultant role in Manchester, salary 32,000 to 38,000 plus uncapped commission, for a 40 person agency specialising in tech and digital roles. Highlight hybrid working and a warm desk with existing client relationships. Keep it under 300 words and end with a clear call to action to apply." The second version gives ChatGPT the real facts a candidate needs to decide whether to apply, so the output needs far less editing.

Mistakes to avoid, and never paste candidate personal data

The single biggest mistake is pasting a candidate's CV, name, email, phone number, salary history or references into ChatGPT to get a summary or screening questions. That data can be stored and used to train future models depending on the settings, and it is a data protection risk your agency cannot take back once it is sent. Strip out names and contact details, or describe the skills and experience in general terms instead.

Other common mistakes: sending outreach messages without adding one real, specific detail about the candidate, which makes the message read as mass produced; trusting ChatGPT's summary of a CV as fact rather than checking it yourself, since it can misread dates, job titles and duration of employment; and using the same three prompts for every role, which produces job ads that all sound the same and stop converting.

I have written more around this on the site: Can You Use ChatGPT Prompts to Avoid AI Detection Tools? Here’s What Happens, Should You Put AI Tools on Your Resume? My Honest Take After Reading Hundreds of Them, How to Have AI Draft a Letter for You (Without It Sounding Like a Robot Wrote It).

This guide is part of my ChatGPT Prompts and Limits: 30 Guides to Fun Prompts, Photo Edits and Daily Caps.

The free prompt builder does the working out for you.

Related: my ai consultant recruitment agencies screening follow up page.

If you're bringing new people into your team, my roundup of the best employee onboarding software shows how to make their first weeks smoother.

Want to contribute a HR tech guest post? My guidelines explain what gets accepted. Experts and practitioners can read my generative AI contributor guidelines before pitching.

If you're hiring for a small team, see my guide to comparing applicant tracking software.

There's more detail in my post on ChatGPT prompts for opticians. Worth reading next: who owns ChatGPT.

Frequently asked questions

Can I paste a candidate's CV into ChatGPT to save time on screening?

You should not paste a full CV, since it usually contains a name, address, phone number, email and sometimes a photo or date of birth. If you want help drafting screening questions, describe the skills and experience level in general terms instead, with no personal identifiers attached.

Will ChatGPT write outreach messages good enough to send without editing?

Not usually. It writes a solid structure and tone, but it will not know the one detail that makes a candidate feel picked rather than mass messaged, such as a specific project they worked on or a shared connection. Add that detail yourself before you send anything.

Does using ChatGPT for job ads hurt how they rank on job boards?

Job boards rank ads mainly on keyword match, salary transparency and how quickly you respond to applicants, not on whether the copy was drafted with help from ChatGPT. A clear, specific ad written with ChatGPT and then edited by a person will usually outperform a vague one written entirely by hand.

How much time does this save a recruiter each week, in real terms?

Recruiters who use structured prompts for job ads, outreach and client emails typically report saving several hours a week, mostly on first drafts rather than final copy. The time saving comes from skipping the blank page, not from skipping the editing.

Is it safe to use ChatGPT for client facing emails?

It is safe for drafting, as long as you remove any client or candidate personal data before you paste in context, and you read the final email back before sending. Treat every draft as a starting point a person needs to check, not a finished message.