- Why most ChatGPT resume prompts fail before you even start
- The base prompt that works
- Prompts for tailoring to each specific job
- Cover letter prompts that don't sound like a template
- The contrarian bit: stop asking ChatGPT to sound impressive
- Formatting and ATS prompts worth running
- Where ChatGPT prompts stop being enough
- A five-step prompt sequence to run start to finish
- Frequently asked questions
The short version: ChatGPT writes dreadful resumes if you ask it to "write me a resume" and surprisingly good ones if you feed it your actual achievements, numbers and the job advert, then ask it to edit rather than invent. The prompts below are built to stop the model guessing and start it working from facts you give it.
I've spent years watching people copy and paste a job title into ChatGPT and expect magic. What they get instead is a resume full of "results-driven professional with a proven track record" nonsense that reads like every other resume a recruiter sees that week. The tool isn't the problem. The prompting is.
Why most ChatGPT resume prompts fail before you even start
ChatGPT doesn't know you did anything unless you tell it. If your prompt is "write a resume for a marketing manager," it has no choice but to invent generic achievements, because it has no real ones to draw on. That's the uncomfortable bit nobody wants to say out loud: the quality of your resume prompt is entirely limited by the quality of the raw material you hand over. Garbage in, polished-sounding garbage out.
So before you open ChatGPT, write down ten things you did in your last job, each one with a number attached. Grew email list from 2,000 to 14,000. Cut onboarding time from six weeks to ten days. Managed a budget of £80,000. Numbers are what separate a resume that gets an interview from one that gets ignored by both the recruiter and the applicant tracking system.
The base prompt that works
Here's a structure I'd use if I were rebuilding a resume from scratch today:
- "Here is my current resume [paste it]. Here is the job advert I'm applying to [paste it]. Rewrite my summary section to match the language and priorities in this job advert, using only facts from my existing resume. Do not invent new achievements."
- "Here are ten bullet points describing what I did in my last role [list them]. Turn each into a resume bullet starting with a strong verb, including the number where I've given one, and keep each under 20 words."
- "Review this bullet point and tell me if it sounds generic or specific: [paste bullet]. If generic, ask me three questions that would help make it specific."
That last one is the prompt most people skip, and it's the one doing the real work. Getting ChatGPT to interrogate you is far more useful than getting it to write for you.
A worked example
Say you run a small bakery supply business and you're applying for an operations manager role at a logistics company. Your old resume says "responsible for supply chain operations." That tells a recruiter nothing. Instead, prompt ChatGPT with: "I managed deliveries to 40 independent bakeries across three counties, reduced late deliveries from 18% to 4% over one year by renegotiating supplier contracts, and cut fuel costs by switching two routes to a different carrier. Turn this into three resume bullets for an operations manager role, each starting with a different strong verb, each under 18 words, with the numbers kept in."
You'll get something like: "Reduced late deliveries by 78% across 40 accounts through supplier contract renegotiation." That's a sentence a hiring manager stops on. It took one prompt and three real facts, not ten minutes of ChatGPT guessing at your job title.
Prompts for tailoring to each specific job
Tailoring is the single biggest lever most people don't pull, mostly because it's tedious. ChatGPT makes it fast rather than optional:
- "Compare this job advert to my resume and list the five keywords or skills mentioned in the advert that are missing from my resume."
- "Of those five, which ones do I have experience with based on what I've told you about my background? Suggest where I could add them."
- "Rewrite my professional summary in 50 words, prioritising the three requirements that appear first in this job advert."
Notice the word in that second prompt. You want ChatGPT catching you before a recruiter does if you're about to claim a skill you don't have. Applicant tracking systems reward keyword matches, but a human interview will expose a lie about Python skills in about ninety seconds.
Cover letter prompts that don't sound like a template
Cover letters are where ChatGPT output is most likely to sound robotic, because the genre itself is already stiff. The fix is giving it something to react to instead of asking it to write from nothing.
- "Here's the job advert [paste]. Here's my resume [paste]. Write a cover letter opening paragraph that mentions one specific thing about the company, taken from their website or recent news, and connects it to one specific achievement of mine."
- "My current cover letter draft is here [paste]. Point out any sentence that could apply to literally any candidate for any job, and suggest how to make it specific to me."
- "Rewrite this closing paragraph so it doesn't say 'I look forward to hearing from you' and instead states clearly what I'd do in the first 30 days in this role."
If you want a faster starting point than a blank page, a decent free cover letter template gives ChatGPT a structure to work within rather than a format to invent, which usually produces a tighter result.
The contrarian bit: stop asking ChatGPT to sound impressive
Every guide on this topic tells you to ask for "powerful, impactful language." Don't. Ask for plain language and specific facts instead. Recruiters read hundreds of resumes a week and they can smell inflated language from across the room. "Spearheaded a transformative initiative" means less to them than "led a four-person team that launched a product six weeks early." Plain and specific beats impressive and vague every single time, and that's the opposite of what most ChatGPT prompt lists tell you to aim for.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
A useful prompt for this: "Rewrite this bullet point removing any adjective that isn't a number or a fact." You'll be shocked how many words get cut, and how much better it reads once they do.
Formatting and ATS prompts worth running
- "List any formatting in this resume that might confuse an applicant tracking system, such as tables, columns, graphics or headers in text boxes."
- "Check this resume for consistent verb tense, consistent date format, and consistent use of bullet points versus paragraphs."
- "Read this resume as if you're a recruiter spending seven seconds on it. What's the first thing you'd notice, good or bad?"
Seven seconds is roughly the average attention a recruiter gives a resume on first pass, so that last prompt is closer to reality than most people assume.
Where ChatGPT prompts stop being enough
Prompting gets you 80% of the way to a strong resume and cover letter. The last 20% is judgement: knowing which achievement matters for this role, knowing when a cover letter should be three paragraphs instead of four, knowing when the tone needs to shift because you're applying to a scrappy startup rather than a bank. No prompt replaces that. If you've tried a handful of tools and still feel stuck, it's worth checking whether a platform is helping or just collecting your card details, which is exactly the kind of question I covered when I looked at whether Resume Now is legit.
If you're applying across multiple roles and want a fuller comparison of what's out there beyond ChatGPT, I've broken down the best AI cover letter generators and how they differ in practice rather than in marketing copy. And if this whole process feels like more than you want to DIY, that's a reasonable thing to outsource to an AI consultant who can set up the workflow once and hand it to you to reuse.
A five-step prompt sequence to run start to finish
- 1. List ten achievements with numbers before opening ChatGPT at all.
- 2. Paste your resume and the job advert together and ask for a keyword gap list.
- 3. Rewrite bullets one at a time, feeding ChatGPT the raw fact and asking for a verb-led sentence under 20 words.
- 4. Ask ChatGPT to strip adjectives that aren't facts or numbers.
- 5. Read the final version aloud. If a sentence sounds like it could belong to anyone, send it back for one more round.
For a deeper look at building prompts specifically around resume structure rather than cover letters, I've written a longer piece on how to write better resumes using AI prompts that goes further into section-by-section rewrites.
For the rest of the series, see the ChatGPT Prompts and Limits: 30 Guides to Fun Prompts, Photo Edits and Daily Caps. To write your own, try the free prompt generator.
Frequently asked questions
Can ChatGPT write a whole resume for me from scratch?
It can produce something resume-shaped, but without your real facts and numbers it will invent generic achievements that won't survive an interview, so always feed it your own material rather than asking it to start from nothing.
Will recruiters know if I used ChatGPT for my cover letter?
They'll know if it reads generically, with vague phrases that could apply to any candidate, but a cover letter built from specific facts and tailored to the job advert reads no differently from one you wrote entirely yourself.
What's the single best ChatGPT prompt for tailoring a resume to a job advert?
Paste both documents and ask ChatGPT to list the skills or keywords in the advert that are missing from your resume, then only add the ones you have experience with.
Should I use the same ChatGPT resume for every job application?
No, tailoring each application to the specific advert using a short gap-analysis prompt takes a few minutes and consistently performs better than sending one generic version everywhere.
Run a business, a brand or a blog? I feature tools, products and expert guest posts for an audience of 15,000. See how it works