ChatGPT Prompts for Resume Writing: 25 Tested Prompts by Section (With Outputs)

Rewrite my professional summary in 35 words, targeting the role of [TITLE] at [COMPANY],
using these accomplishments:
1. [accomplishment with metric]
2. [accomplishment with metric]
3. [accomplishment with metric]

Constraints: lead with a quantified result, name the function (e.g., "data engineer"
or "B2B SaaS PM"), do not use the phrases "results-driven," "passionate," or
"dynamic," do not invent any metric I have not provided. Output one paragraph only.

Sample output:

Data engineer with 6 years building Snowflake pipelines for fintech analytics teams. Cut warehouse spend 34% across two migrations, shipped 12 production ELT flows, and mentored 4 analysts on dbt testing standards.

Fix to apply: Confirm the 34%, 12, and 4 are real numbers from your record, not numbers ChatGPT inferred. Tighten the noun "fintech analytics teams" to the exact subsector on the job posting if it is more specific.

Generate three professional summary variants for a career change from
[PRIOR ROLE] to [TARGET ROLE]. Each variant should be 40 to 55 words.

Variant A leads with transferable skills.
Variant B leads with a recent project that bridges the two domains.
Variant C leads with the credential or training that supports the pivot.

My prior role: [paste prior bullets].
My target role JD: [paste JD].
Recent bridging project: [describe it, with metric].
Bridging credential or course: [name it].

Do not use "passionate," "excited," or "transitioning." State the pivot as
a current fact, not an aspiration.

Sample output:

Variant B: Customer success lead moving into product management on the back of an 8-month embed with the platform team, where I shipped 3 onboarding features that lifted activation 19%. Strong in user-research synthesis, JIRA roadmapping, and stakeholder writing.

Fix to apply: Pick one variant; do not paste all three into the resume. Verify the 19% lift and the 8-month duration. Replace "Strong in" with concrete tool names from the target JD.

Tighten this summary to remove all filler and AI-tell phrases. Keep every concrete
fact, number, and proper noun. Cut adjectives that do not carry weight. Target
length: 45 words or less.

Banned phrases: results-driven, dynamic, passionate, proven track record,
detail-oriented, team player, seasoned, motivated self-starter, leveraged, synergy,
spearheaded, multifaceted, robust, cutting-edge, value-add.

Input summary: [paste your current summary].

Sample output:

Senior product designer with 7 years across fintech and healthtech. Shipped 14 production features, including a claims dashboard that reduced support tickets 22%. Comfortable in Figma, ProtoPie, and a weekly research cadence with 6 to 8 customers.

Fix to apply: Verify the 14, 22%, and 6 to 8 numbers are accurate. Replace "Comfortable in" with "Daily in" if the tools are core to your workflow; recruiters read "comfortable" as low confidence.

Convert this LinkedIn About section to a resume professional summary.

Differences to apply:
1. Resume summary is third-person voice (implied subject), not first-person.
2. Max 50 words.
3. Lead with role and years of experience, not a story.
4. Cut every personal anecdote.
5. Preserve every metric and proper noun.

LinkedIn About: [paste it].

Sample output:

Marketing manager with 5 years in B2B SaaS demand generation. Built a paid funnel from $0 to $1.2M ARR contribution in 18 months at a Series B startup. Strong in HubSpot, Demandbase, and SQL-driven campaign reporting.

Fix to apply: Drop "Strong in" if you have one tool you lead with daily; promote it to the action verb instead, for example "Runs HubSpot and Demandbase as primary stack."

Rewrite this bullet using the Action + Tool + Outcome structure with a quantified
result. Action is a past-tense verb that is not a banned filler. Tool is the
specific platform, framework, or method. Outcome is a number, percentage, or
magnitude tied to a business goal.

Banned verbs: leveraged, utilized, assisted with, worked on, helped, was
responsible for.

Original bullet: [paste bullet].
Real metric I can claim: [number and what it measured].

Sample output:

Rebuilt the order-status microservice in Go with Redis caching, cutting p95 response time from 480ms to 95ms and shaving 8% off monthly Kubernetes spend.

Fix to apply: Verify both numbers (480 to 95ms, 8%). If you only have the latency improvement, drop the spend claim rather than carry an unsourced number into a recruiter conversation.

Convert this responsibility-style bullet into an achievement-style bullet.

Definition: a responsibility bullet describes what you were assigned. An
achievement bullet describes what changed because you did the work.

Requirements:
- Lead with a past-tense action verb (not "responsible for").
- Name the outcome in concrete terms (number, percentage, time saved, revenue,
  defects avoided, customers served).
- Keep under 28 words.

Original bullet: [paste it].
What actually changed because of my work: [describe it, even if rough].

Sample output:

Owned weekly close for a 9-entity Salesforce-Netsuite stack, closing books in 4 days versus the prior 7 and clearing 100% of audit findings two quarters running.

Fix to apply: Confirm the "9-entity," "4 days versus 7," and "100% of audit findings" against your records. Numbers are the part recruiters and interviewers will probe.

Generate three variants of this bullet at different seniority framings: IC, lead,
and manager.

IC framing: emphasizes individual delivery and craft.
Lead framing: emphasizes setting direction for 2 to 5 peers without formal authority.
Manager framing: emphasizes headcount, hiring, and outcomes across a team.

Original bullet: [paste].
Real scope I can defend in interview: [headcount, budget, scope].

Sample output:

Lead framing: Set the QA strategy for a 4-engineer payments squad, introduced contract testing with Pact, and reduced regression escapes 41% over two release cycles.

Fix to apply: Use only the framing that matches the role you are applying for; do not stack all three. Overclaiming "manager" without real headcount is a fast way to lose trust in a screen.

Strip industry jargon and convert into language a recruiter outside [INDUSTRY]
would understand. Preserve every metric and every proper-noun tool name; recruiters
still need to keyword-match those. The goal is to remove insider phrasing that
hides the impact.

Original bullet: [paste].
Industry: [name it].

Rule: if you remove a jargon phrase, replace it with a plain-English description
in the same number of words or fewer.

Sample output:

Cut prior-authorization turnaround from 6 days to 2 days for a 14-clinic network by automating insurer eligibility checks in Epic, freeing 18 staff hours per week.

Fix to apply: Test the bullet on someone outside your function. If they cannot describe the impact in one sentence after reading it, simplify a noun.

Add metrics to this bullet using these performance numbers. Do not invent any
number I have not provided. If a number does not fit, leave it out rather than
forcing it in.

Original bullet: [paste].
Numbers I can claim, with the source:
- [number 1, what it measured, where it came from]
- [number 2, what it measured, where it came from]
- [number 3, what it measured, where it came from]

Output one bullet, 24 words or less.

Sample output:

Shipped a self-serve onboarding flow that lifted activation from 38% to 57% across 12,400 monthly signups and cut implementation tickets 31%.

Fix to apply: Cross-check each number against the source you cited in the prompt. ChatGPT occasionally rounds or transposes digits.

Detect and remove AI-tell phrases from these bullets. For each removal, return:
1. The original sentence.
2. The AI-tell phrase found.
3. The rewrite.

Phrases to flag: leveraged, leveraging, results-driven, passionate, dynamic,
proven track record, robust, cutting-edge, value-add, mission-critical, best
practices, holistic, synergy, spearheaded, multifaceted, world-class, top-tier,
solutions-oriented, detail-oriented, results-oriented.

Bullets: [paste 5 to 10 bullets].

Sample output:

Original: Leveraged data-driven insights to drive cross-functional alignment. Phrase: leveraged, data-driven, cross-functional alignment. Rewrite: Used product analytics to land a single Q3 roadmap with engineering, design, and revenue.

Fix to apply: Accept the rewrite only if the rewrite is also true. ChatGPT will sometimes invent a tool or outcome to plug a gap; cut that part before pasting back into the resume.

Extract a skills section from this resume and this job description, prioritizing
the JD's terminology. Output a flat comma-separated list, 12 to 18 items.

Priority order:
1. Exact-match terms that appear in both the JD and my experience.
2. JD terms I have real experience with, even if my resume currently phrases
   them differently.
3. Core tools I use weekly but the JD did not mention.

Do not include skills I cannot defend in an interview.

Resume: [paste].
JD: [paste].

Sample output:

Python, SQL, dbt, Snowflake, Airflow, AWS (S3, Lambda, Glue), Terraform, Git, Looker, data modeling, Kimball dimensional design, stakeholder requirements gathering, CI/CD for analytics, on-call rotation, mentoring junior engineers.

Fix to apply: Remove any item you would not be comfortable being grilled on in a technical interview. A short, defensible list beats a long, padded one.

Split these skills into three buckets and return as a short structured list:
1. Hard skills (technical, measurable, teachable).
2. Soft skills (behavioral, interpersonal).
3. Tools and platforms (named products and frameworks).

Rules:
- Each bucket maxes at 8 items.
- A single skill may appear in only one bucket.
- Drop any skill that overlaps with a different bullet on the resume already.

Input list: [paste].

Sample output:

Hard: SQL, data modeling, A/B test design, regression analysis, dashboard design. Soft: stakeholder communication, technical mentoring, written briefs. Tools: Snowflake, dbt, Looker, Mode, Hex, Sigma, Git, Jira.

Fix to apply: Many resumes do not need a separate soft-skills row; if the bullets already demonstrate communication and mentoring, cut the soft bucket and reclaim the space for hard skills.

Suggest 5 additional skills I should add based on this job description gap analysis.

For each suggestion, return:
1. The skill, phrased exactly as the JD phrases it.
2. The frequency it appears in the JD.
3. A check on whether my resume already implies it (yes/no).
4. A note on whether I should add it only if I can defend it.

Do not suggest skills the JD does not mention. Do not pad the list.

My resume: [paste].
JD: [paste].

Sample output:

1. "Experimentation platform ownership" (3 mentions, not on resume, add only if I owned LaunchDarkly or Statsig). 2. "Vendor management" (2 mentions, partly implied, safe to add as an explicit skill).

Fix to apply: Treat every "add only if I can defend" as a real gate. Adding a JD phrase you cannot defend tanks the screen as soon as it comes up in conversation.

Convert these freeform skill mentions in my experience bullets into a clean
skills section. The goal is to surface tools and methods that the ATS will
miss when they are buried mid-sentence.

Rules:
- Pull only proper-noun tools and named methodologies.
- Deduplicate.
- Order by frequency across bullets (most-mentioned first).
- Keep the list under 15 items.

Bullets: [paste].

Sample output:

Salesforce, HubSpot, Marketo, SQL, Tableau, Snowflake, Segment, dbt, Looker, GTM Bowtie framework, ABM, account scoring, attribution modeling.

Fix to apply: Drop frameworks like "GTM Bowtie" if they are not in the target JD; in-house language buys nothing on a generic ATS pass.

Identify the top 15 ATS keywords in this job description, ranked by frequency.

For each keyword, return:
- The exact phrase as written in the JD.
- Its raw count.
- A category tag: tool, methodology, soft skill, domain, certification.

Ignore generic words (work, team, role). Combine close variants only when
they are the same concept (e.g., "Python" and "Python 3").

JD: [paste].

Sample output:

1. Salesforce (7, tool). 2. quota attainment (4, domain). 3. consultative selling (3, methodology). 4. MEDDIC (3, methodology). 5. enterprise (3, domain). 6. forecasting (2, soft skill).

Fix to apply: Confirm the counts by searching the JD for each term. The categorization is judgment, not parser truth; recategorize where it helps your skills bucket layout.

Compare my resume against this JD and list missing keywords.

Return three columns:
1. Keyword (exact JD phrasing).
2. JD frequency.
3. Likely placement on resume (skills, bullet, summary).

Only flag keywords that are concrete (tools, methodologies, certifications,
domain nouns). Skip filler verbs and generic adjectives. Do not suggest a
keyword I cannot defend; if you are unsure, label it "verify before adding."

Resume: [paste].
JD: [paste].

Sample output:

1. "Terraform" (4, skills, verify before adding). 2. "incident command" (3, bullet on on-call work). 3. "SOC 2 readiness" (2, summary).

Fix to apply: Place each missing keyword into the column it actually belongs to in your resume. A bullet-level keyword should live inside a real accomplishment, not a comma list.

Rewrite this section to include these keywords naturally without keyword stuffing.

Rules:
- Each keyword should appear once unless the original work genuinely justifies repeat use.
- Bullets must remain achievement-style (Action + Tool + Outcome).
- Do not add a number that was not already in the original.
- Reject any keyword that does not fit; return it in a "could not place" list at the bottom.

Section: [paste resume section].
Keywords to weave: [comma-separated list from Prompt 16].

Sample output:

Bullet rewrite: Led incident command for 11 P1 outages across the payments stack, reducing mean time to resolution from 47 to 19 minutes through Terraform-managed runbook automation. Could not place: SOC 2 readiness (no real exposure).

Fix to apply: Honor the "could not place" output. Forcing a keyword into a bullet you cannot defend is more damaging than skipping it.

Detect keyword stuffing in this bullet and rewrite for natural reading.

Stuffing definitions:
1. The same proper-noun keyword appears 3+ times in one bullet.
2. A list of nouns runs more than 5 items without verbs.
3. A bullet reads like a tag cloud rather than a sentence.

For each detection, return: the bullet, the stuffing pattern, the rewrite.
Preserve all real metrics. Output the rewrite at 26 words or less.

Bullets: [paste].

Sample output:

Original: Used Salesforce, Salesforce CPQ, and Salesforce Service Cloud to manage Salesforce data. Pattern: repeated proper noun (4x). Rewrite: Owned the full Salesforce stack (CPQ, Service Cloud) for a 60-rep org, lifting quote-to-close speed 28%.

Fix to apply: Re-check that the rewrite preserves the keyword coverage you need. Pulling stuffing out without losing the right ATS signal is the harder half.

Draft a 3-paragraph cover letter for this role using my resume and this JD.

Paragraph structure:
1. Why this role and company specifically (use a real fact from the company,
   not a generic compliment). 2 sentences.
2. Two of my strongest, most relevant accomplishments from the resume, with
   metrics. 3 sentences.
3. A direct ask for a conversation. 1 sentence.

Total under 250 words. Do not use "I am excited," "I am passionate," or
"results-driven." Do not invent facts about the company.

Resume: [paste].
JD: [paste].
Real company fact I will provide: [name a launch, hire, funding round, podcast quote, or news item with the source].

Sample output:

Your Q1 platform refactor and the move to a single billing API caught my attention; consolidating that surface is the exact problem I have been shipping against for the last two years.

Fix to apply: Verify the company fact is real and current. ChatGPT will sometimes hallucinate product launches; if it returns a fact you did not give it, delete it.

Write a 6-line outreach LinkedIn DM to a recruiter for this role.

Rules:
- Line 1: greeting + one specific reason for reaching out to this recruiter (a
  role they posted, a hire they announced, a comment they wrote).
- Lines 2 to 3: one accomplishment from my resume that maps to the role, with
  a metric.
- Line 4: one sentence on what I am open to (full-time, contract, location).
- Line 5: a soft ask for a 15-minute call.
- Line 6: thanks + name.

Total under 90 words. No emojis. No "circling back."

Resume: [paste].
JD or role title: [paste].

Sample output:

Hi Priya, saw the Sr. Data Engineer role you posted last week. I most recently shipped a dbt + Snowflake migration that cut warehouse spend 34% across a 200-model graph. Open to full-time, remote-US. Worth 15 minutes this week or next? Thanks, Sam.

Fix to apply: Rewrite the greeting line in your own voice; ChatGPT's openings are usable but slightly stiff. Keep the metric exactly as drafted if it matches your numbers.

Generate three follow-up email variants for the scenario: [post-interview,
post-application, post-rejection]. Each variant is a complete email body.

Variant A: short and warm (under 70 words).
Variant B: short and substantive (one new fact or artifact attached).
Variant C: re-engagement at 10 to 14 day mark with a specific question.

Rules: subject lines are 6 words or fewer, no "just checking in," no "wanted to
circle back," no emojis. Sign-off uses my first name only.

Role and stage: [describe].
What I learned in the prior touchpoint: [paste].

Sample output:

Variant B subject: Quick artifact from our chat. Body: Thanks again for Thursday. You mentioned the migration timeline for the billing system; here is a 1-pager I wrote last quarter on a similar consolidation with the rollback plan we used.

Fix to apply: Only send Variant B if the artifact actually exists. Promising and not sending kills the thread faster than no follow-up at all.

Audit this resume for AI-tell phrases and flag every instance.

For each instance, return:
1. The exact sentence.
2. The phrase that flagged it.
3. A one-sentence replacement that preserves the underlying fact.

Phrase list: leveraged, results-driven, passionate, dynamic, robust, cutting-edge,
synergy, spearheaded, value-add, mission-critical, best-in-class, holistic,
seasoned, motivated self-starter, detail-oriented, world-class, top-tier,
proven track record, multifaceted, solutions-oriented.

Resume: [paste full resume].

Sample output:

Sentence: Leveraged cutting-edge analytics to drive results-driven outcomes. Flags: leveraged, cutting-edge, results-driven. Replacement: Built a churn model in Python that lifted retention 11% on a 40,000-account base.

Fix to apply: Treat the replacement as a suggestion. If the replacement adds a number you have not verified, swap in the real number before pasting it back into your resume.

Compare this resume against this job description and grade alignment from 0 to 100
with specific gaps.

Return:
1. Score (0 to 100).
2. Top 5 strengths (exact JD-resume matches).
3. Top 5 gaps (JD requirements not present or under-evidenced).
4. Three highest-leverage edits to close the gap.

Be conservative. Do not credit a match unless the resume actually demonstrates it.

Resume: [paste].
JD: [paste].

Sample output:

Score 68/100. Strengths: dbt (3 bullets), Snowflake (2 bullets), stakeholder writing. Gaps: experimentation platform ownership, on-call rotation, vendor management. Highest-leverage edit: add a bullet on the LaunchDarkly migration.

Fix to apply: Use the score as direction, not gospel. ChatGPT's 0 to 100 is a rough heuristic; for a real parse, run the resume through our free ATS resume checker after applying the edits.

Detect inconsistent verb tenses across all experience bullets.

Rules:
1. Past roles must use past tense throughout.
2. The current role uses present tense; one-time achievements within the
   current role can use past tense if clearly bounded.
3. Mixing tense within a single role flags as an error.

For each error, return: the role, the bullet, the offending verb, the correction.

Resume: [paste].

Sample output:

Role: Acme Corp (2022 to 2024). Bullet: Manage the QA strategy for a 4-engineer payments squad. Offender: "Manage" in a closed-date past role. Correction: "Managed."

Fix to apply: Apply every flagged correction; tense mistakes are the most-common machine-detected error and the easiest to fix in one pass.

Flag any unsubstantiated claims or numbers that read as fabricated in this resume.

Heuristics for flagging:
1. Round numbers without a tool (e.g., "increased revenue by 30%" with no
   source named).
2. Vague magnitudes ("significantly improved," "drastically reduced").
3. Claims that would be hard to defend in interview given the role's scope.

Return: the bullet, the flag reason, a follow-up question I should be able to
answer in interview.

Resume: [paste].

Sample output:

Bullet: Drove a 30% revenue lift across the portfolio. Flag: round number, no tool or method named, scope unclear. Interview question: "Which channel and what baseline?"

Fix to apply: Either back the claim with a defensible source or cut it. Carrying a number you cannot defend into a real interview is worse than leaving the bullet softer.