ChatGPT Prompts for Sales (2026): 40 Tested, Copy-Paste
These are 40 ChatGPT prompts for sales that hold up in a real pipeline — written for SDRs, AEs, and sales managers, organized by deal stage from first research to closed-won ops. Every prompt is free, ungated, and copyable in one click, with ALL-CAPS variables to fill and constraints that keep the output from sounding like a machine wrote it. Most run unchanged in Claude and Gemini.
You cannot personalize what you have not read. These eight prompts do the reading — account briefs, trigger events, buying committees — so your first touch lands with proof of homework. If top-of-funnel is your entire job, the dedicated SDR prospecting prompts collection goes deeper still.
01
Account Snapshot Builder
Use this when you have twenty minutes of raw research material and need a one-page brief before writing a single word of outreach.
SDR
Prospecting
ChatGPT
You are a B2B sales research analyst. I sell [YOUR PRODUCT] to [PERSONA] in [INDUSTRY]. Below I will paste raw research on [PROSPECT COMPANY]: website copy, a press release, an annual-report excerpt, or my own notes.
Build a one-page account snapshot with exactly these five sections: 1) What they sell and to whom, in two sentences. 2) Three recent signals that matter to a seller (funding, hires, product moves, leadership changes), each with the source line it came from. 3) The two business priorities they state most loudly, quoted directly. 4) One plausible problem [YOUR PRODUCT] could touch, phrased as a hypothesis, not a fact. 5) Three opening angles for a first touch, one sentence each.
Rules: use only what appears in the pasted material — if a section lacks evidence, write NOT FOUND IN SOURCE instead of guessing. Mark every inference with "(inference)". Plain language, no marketing adjectives. Total length under 300 words.
Research material: [PASTED RESEARCH MATERIAL]
Use this when a new patch of accounts just landed in your name and you need a defensible working order by Friday.
SDR
Prospecting
ChatGPT
Act as a sales operations analyst helping me prioritize a new territory. My ideal customer profile: [IDEAL CUSTOMER PROFILE — industry, size, tools, buying triggers]. I sell [YOUR PRODUCT] at a typical deal size of [DEAL SIZE].
I will paste a list of accounts with whatever fields I have (name, industry, headcount, region, notes). Sort every account into three tiers: Tier 1 — work this week. Tier 2 — nurture this month. Tier 3 — deprioritize.
Output a table with columns: Account, Tier, One-line reason, Missing data that would change the tier. After the table, list the three data points that would most improve this ranking if I collected them.
Rules: never invent firmographics I did not provide. If a field is blank, score conservatively and say so in the reason column. Reasons must reference my ICP criteria by name, not vibes. If my ICP description is too vague to score against, ask me up to 3 questions before ranking anything.
Account list: [ACCOUNT LIST]
Use this when a news alert fired on a target account and you want the angle before it goes stale.
SDR
Prospecting
ChatGPT
You are an outbound strategist. I will paste one or more news items about [PROSPECT COMPANY] — funding, a leadership hire, layoffs, an acquisition, a product launch, or an earnings note. I sell [YOUR PRODUCT] to [PERSONA].
For each item, produce: 1) The event, in one neutral sentence. 2) Why it plausibly creates pressure on [PERSONA] within 90 days — two sentences, grounded in the event itself. 3) An outreach angle written as one sentence I could open an email with, referencing the event without congratulating them. 4) A risk note: one way this angle could misread the situation.
Rules: no flattery openers ("Congrats on..."), no assumed pain the event does not support, no more than 80 words per item. If an item is too thin to support an angle, write SKIP — no seller-relevant pressure — and move on.
News items: [PASTED NEWS ITEMS]
Use this when you are entering a new vertical and want pain hypotheses to test in discovery instead of guessing live on a call.
SDR / AE
Prospecting
ChatGPT
Act as a former [PERSONA] with ten years in [INDUSTRY] who now advises sales teams. I sell [YOUR PRODUCT].
Generate five pain hypotheses this persona likely lives with that [YOUR PRODUCT] could plausibly address. For each: 1) The pain, stated the way the persona would describe it to a peer — their words, not vendor words. 2) The upstream cause. 3) What it costs them, described qualitatively (time, risk, credibility — no invented dollar figures). 4) A discovery question that tests whether this pain is real for a specific account, phrased so a "no" is easy and informative.
Rules: hypotheses must be genuinely distinct — no rephrasing one pain five ways. Rank them by how likely they are to be top-of-mind this quarter and defend the ranking in one sentence each. Plain language throughout. If you need context about my product's capabilities to do this well, ask me up to 3 questions before answering.
Use this when you have a list of names and titles from the account and need to know who actually decides.
AE
Prospecting
ChatGPT
You are a deal strategist. I will paste job titles (and notes, where I have them) for people at [PROSPECT COMPANY]. I sell [YOUR PRODUCT] at roughly [DEAL SIZE] deal size.
Map a likely buying committee. For each person, assign one probable role — economic buyer, champion candidate, technical evaluator, end user, blocker risk, or unclear — with a one-sentence rationale grounded in the title. Then: 1) name the two people to contact first and why; 2) sketch the most likely approval path for a deal this size as a simple arrow chain of titles; 3) list who is missing — roles a deal like this usually needs that I have no name for yet.
Rules: this is triage from titles, so state confidence (high / medium / low) for every person and never present a guess as a fact. Format as a table followed by three short paragraphs.
Titles and notes: [JOB TITLES LIST]
Use this when the account's careers page is the most honest document they publish — because it usually is.
SDR
Prospecting
ChatGPT · also strong on Claude
Act as a competitive intelligence analyst. Job postings are the most candid thing a company publishes: they name tools, team structures, and unsolved problems. I will paste one or more postings from [PROSPECT COMPANY]. I sell [YOUR PRODUCT] to [PERSONA].
Extract: 1) Tools and systems named or implied, with the exact line that implies each. 2) Team shape — what this role reports into and what that suggests about the org. 3) Stated problems — responsibilities that read like cleanup ("own the migration", "bring order to..."), quoted. 4) One outreach angle connecting a finding to [YOUR PRODUCT], written as a two-sentence email opener that quotes or closely paraphrases the posting.
Rules: every claim must trace to quoted text from the posting — include the quote. No inferences more than one step from the text. If the posting is generic boilerplate with no signal, say so plainly instead of manufacturing insight.
Postings: [JOB POSTING TEXT]
Use this when the prospect already pays your competitor and "we already have a tool" is coming.
AE
Prospecting
ChatGPT
You are a displacement-deal strategist. [PROSPECT COMPANY] currently uses [COMPETITOR]. I sell [YOUR PRODUCT]. My honest differentiators, in my own words: [YOUR DIFFERENTIATORS].
Build a displacement brief: 1) Three moments when a company genuinely reconsiders an incumbent tool (renewal window, ownership change, missed expectation) and the signal that would tell me each moment is near. 2) The switching costs my prospect will weigh, listed honestly — including the ones that favor [COMPETITOR]. 3) Two discovery questions that surface dissatisfaction without trashing the incumbent. 4) A 60-word first-touch email positioned around curiosity about their current setup, never an attack on [COMPETITOR].
Rules: use only the differentiators I provided — do not invent competitor weaknesses or features for either product. Never disparage [COMPETITOR] by name in the drafted email. If my differentiators are too thin to support a displacement motion, tell me that directly and name the evidence I should gather first.
Use this when the meeting is booked, the research is done, and you need questions that prove you did it.
AE
Prospecting
ChatGPT · also strong on Gemini
You are a sales coach preparing me for a first meeting with [PROSPECT COMPANY]. My goal for this call: [MEETING GOAL]. I will paste my research notes below.
Write ten questions that could only be asked by someone who read this research — each must reference a specific fact from my notes. Organize them: three openers that establish credibility fast, four middle questions that test my pain hypotheses, and three that surface decision process and timing without sounding like a checklist.
For each question, add one line: what a strong answer sounds like versus a weak one, so I can qualify in real time.
Rules: no generic questions that would work on any company ("what keeps you up at night"). Questions must be open-ended — nothing answerable with yes or no except one deliberate closing confirmation. Under 25 words per question. If my notes are too thin to support ten grounded questions, write fewer and tell me what research to add.
Research notes: [PASTED RESEARCH NOTES]
Eight prompts built to earn the reply, not the delete: first touches, sequences, call openers, and the two notes most reps get wrong — the revival and the goodbye. If cold email is your entire week, the full cold email prompts collection covers that stage wall to wall.
09
The 90-Word First Touch
Use this when it is your first email to a cold prospect and you get one shot at nine seconds of attention.
SDR
Cold outreach
ChatGPT
Write a cold email from me, [YOUR NAME], [YOUR ROLE] at [YOUR COMPANY], to [PROSPECT NAME], [PROSPECT TITLE] at [PROSPECT COMPANY].
The one piece of research I am building this around: [TRIGGER OR OBSERVATION]. What I sell, in plain words: [YOUR PRODUCT AND THE PROBLEM IT ADDRESSES].
Structure: one observation-led opening line that proves I did homework (never "I hope this finds you well", never congratulations), two sentences connecting the observation to a problem their role owns, one sentence on how we address it without listing features, and a low-friction interest question — not a meeting ask.
Rules: 90 words maximum. Subject line of five words or fewer. No exclamation marks, no marketing adjectives, no "quick question" subject. Write at an eighth-grade reading level and mention my product at most once. Give me two variants: one leading with the problem, one leading with the observation. If any bracketed input is missing, ask me for it before drafting.
Use this when one email will not carry the account and touches two and three should be planned before touch one sends.
SDR
Cold outreach
ChatGPT
Design a three-email cold sequence to [PROSPECT TITLE] at [PROSPECT COMPANY] about [YOUR PRODUCT]. Research I have: [RESEARCH NOTES]. Spacing: day 1, day 4, day 9.
Each email must take a genuinely different angle — not the same pitch re-sent. Email 1 opens with my strongest research-based observation. Email 2 shares a specific way teams like theirs approach the problem, framed as "how peers handle this" — no invented customer names, no invented numbers. Email 3 is a short, warm close-out that states plainly I will stop emailing and leaves the door open.
Per email: subject under six words, body under 80 words, one idea only, one question maximum. Banned phrases: "just following up", "bumping this", "circling back", "checking in", and any apology for emailing. No exclamation marks.
Output as: SUBJECT / BODY / one line explaining why this angle follows from the previous touch. If my research notes are too thin, ask me up to 3 questions before drafting.
Use this when the body is written and the subject line will decide whether anyone ever reads it.
SDR
Cold outreach
ChatGPT
You are an email deliverability and copy specialist. Here is my cold email body: [EMAIL BODY]. The recipient: [PROSPECT TITLE] at [PROSPECT COMPANY].
Generate twelve subject lines in four styles, three of each: 1) Plain internal-memo style, as if a colleague sent it. 2) Specific-detail style, quoting a concrete fact from the email body. 3) Question style — a question the email actually answers. 4) Blunt-offer style — what I am proposing, stated flatly.
For each line: give a one-sentence rationale tied to this recipient's inbox reality, flag any word likely to trip spam filters, and note whether it overpromises relative to my body copy — a subject the body cannot cash is a delete.
Then pick your top two and defend the choice in three sentences total.
Rules: five words or fewer per line, no clickbait, no fake reply prefixes, no title case, no emoji. If the email body I pasted is weak enough that no subject line saves it, say so first.
Use this when you have fifteen seconds after "hello" and the next line decides the call.
SDR
Cold outreach
ChatGPT
Script my cold call opening to [PROSPECT TITLE] at [PROSPECT COMPANY]. I sell [YOUR PRODUCT]. My research nugget: [RESEARCH NUGGET].
Write: 1) A 15-second opener — permission-based, names the reason for my call using the research, and ends with a question that invites a real answer, not "did I catch you at a bad time". 2) Three response branches, each with a two-line reply: "I'm busy" (offer a 30-second version, then earn more), "we already handle that" (curiosity about how, never combat), and "not interested" (one graceful probe, then a polite exit that leaves the door open). 3) A voicemail fallback under 20 seconds that gives one concrete reason to call back — write it with the pauses marked.
Rules: conversational spoken English, contractions welcome, zero jargon, nothing that sounds read aloud. Mark the words to stress in caps. Total script under 220 words.
Use this when the prospect lives on LinkedIn and email is going nowhere.
SDR
Cold outreach
ChatGPT
Write a two-message LinkedIn approach to [PROSPECT NAME], [PROSPECT TITLE] at [PROSPECT COMPANY]. The context I have: [WHY THIS PERSON — a post they wrote, a mutual group, a company event]. I sell [YOUR PRODUCT].
Message 1 — the connection request note, under 280 characters: reference the specific context, state plainly why I want to connect, no pitch.
Message 2 — sent three days after acceptance, under 60 words: thank them without groveling, make one observation connecting their world to a problem I work on, and ask one specific question they can answer from their phone in a single line. No links, no attachments, no meeting ask yet.
Rules: it must read like a person typing on a phone, not a sequence tool. Banned: "synergy", "touch base", "pick your brain", "I'd love to". Give two tone variants of each message: straightforward and slightly playful. If my context is too weak to justify the outreach, say so instead of drafting.
Use this when they said no months ago and something real has changed since.
AE
Cold outreach
ChatGPT
Write a re-approach email to [PROSPECT NAME] at [PROSPECT COMPANY], who evaluated [YOUR PRODUCT] and chose [WHAT THEY DID INSTEAD — a competitor, building in-house, doing nothing] about [TIME SINCE] ago. Their stated reason at the time: [LOSS REASON]. What has changed since: [WHAT CHANGED — a new capability, a new pricing model, their new funding, a leadership change].
Structure: acknowledge the earlier decision in one line without re-litigating it, name the specific change and why it touches their original objection, then ask one question about how their current approach is holding up — curious, not smug. No "I told you so" energy anywhere. Do not pitch a demo; the ask is a conversation.
Rules: under 100 words. Subject line references the earlier evaluation in four words or fewer. Plain language, no urgency phrasing. Give me two versions: one assuming we parted warmly, one assuming we did not. If [WHAT CHANGED] is weak, tell me to wait instead of sending.
Use this when it is five touches, zero replies, and one honest note is left before you stop.
SDR
Cold outreach
ChatGPT · also strong on Claude
Write the final email in a dead outreach thread to [PROSPECT NAME], [PROSPECT TITLE] at [PROSPECT COMPANY]. What I sent before, in one line: [PREVIOUS TOUCHES SUMMARY]. I sell [YOUR PRODUCT].
The goal is an honest close-out a human would respect: acknowledge the silence without guilt-tripping, restate in one sentence the single problem I believed was worth their time, state plainly that I will stop reaching out, and leave one specific no-cost pointer they can use without ever talking to me — a question to ask their own team or a number to check, derived from [THE PROBLEM], not a gated asset.
Rules: under 75 words. Banned: "break up with you", "closing the loop", "should I stay or should I go", and any cleverness that has appeared in a thousand sequence templates. No sad tone, no passive aggression, no pressure phrasing. Two variants: one plain, one with a single dry line of wit. Subject: three words, lowercase.
Use this when someone in your network knows the person you are trying to reach.
SDR / AE
Cold outreach
ChatGPT
Draft an introduction-request email to [CONNECTOR NAME], who knows [TARGET NAME], [TARGET TITLE] at [TARGET COMPANY]. My relationship to the connector: [RELATIONSHIP]. Why I want the intro, honestly: [REASON].
Write two parts. Part 1 — the ask to my connector, under 80 words: state plainly what I am asking, make declining easy ("if this is awkward, ignore this and we're fine"), and promise I will not make them look bad. Part 2 — the forwardable blurb: a three-sentence paragraph they can paste, written in third person about me, that tells [TARGET NAME] exactly why a conversation might be worth twenty minutes, referencing [SPECIFIC RELEVANCE TO TARGET].
Rules: the blurb must be about the target's problem, not my product — my product gets half of one sentence. No superlatives about me, no adjectives a stranger would roll their eyes at. Plain language throughout. If [SPECIFIC RELEVANCE TO TARGET] is empty, ask me for it before writing anything.
The deal is won or lost in the questions. These eight turn calls and transcripts into evidence — MEDDICC gaps, champion signals, cost math — so your pipeline runs on what the buyer said, not on how the call felt.
17
Discovery Question Ladder
Use this when you are prepping the first real discovery call and need questions that dig instead of drift.
AE
Discovery
ChatGPT
Build a discovery question ladder for my call with [PROSPECT TITLE] at [PROSPECT COMPANY] about [YOUR PRODUCT]. What I know so far: [CONTEXT NOTES].
Structure it in four rungs, three questions each. Rung 1 — Situation: current-state questions I could not answer from research; skip anything a search engine answers. Rung 2 — Problem: questions that surface friction in their own words. Rung 3 — Impact: what the friction costs, who feels it, and what happens if nothing changes this year. Rung 4 — Decision: how change actually gets bought here — budget shape, evaluators, past attempts.
For each question, add a one-line "listen for" note: the phrase or signal that means dig deeper here. Then write the one universal follow-up I should use whenever an answer is vague — their version of "tell me more".
Rules: open-ended questions only, under 22 words each, no leading questions that fish for compliments about my product, and phrasing I could actually say aloud without sounding like a form.
Use this when the call ended, the transcript exists, and the pipeline review is Thursday.
AE
Discovery
ChatGPT · also strong on Claude
You are a MEDDICC deal inspector. I will paste a discovery call transcript. Analyze it against every element: Metrics, Economic buyer, Decision criteria, Decision process, Identify pain, Champion, Competition.
For each element output four lines: STATUS — evidenced, partial, or missing. EVIDENCE — the strongest supporting quote from the transcript, verbatim. GAP — what specifically remains unknown. NEXT MOVE — the single question or action that closes the gap, written as words I could actually say on the next call.
Then a summary: the three biggest risks in this deal, ranked, plus the one thing the transcript shows going genuinely well.
Rules: quote only text that appears in the transcript — never present a paraphrase as a quote. If the transcript is thin on an element, resist filling the gap with optimism; "missing" is a legitimate status. Keep the whole output under 500 words so it pastes cleanly into my CRM notes.
Transcript: [CALL TRANSCRIPT]
Use this when someone on the call was friendly — because friendly is not a champion.
AE
Discovery
ChatGPT
Analyze this call transcript for champion signals around [CONTACT NAME]. A real champion has three properties: a personal stake in solving the problem, influence with the economic buyer, and demonstrated willingness to spend social capital for us.
For each property, find evidence for or against — quoting the transcript verbatim for every claim. Distinguish champion behavior (offered to set up a meeting with their VP, used "we" when talking about implementation, pushed back on their own team's status quo) from mere friendliness (agreed a lot, complimented the demo, said "this is cool").
Output: 1) A verdict — champion, potential champion, friendly contact, or unclear — with a confidence level. 2) The evidence table. 3) Three specific tests I can run in the next two weeks to develop or confirm them, each phrased as an ask I could actually make ("would you be open to walking your CFO through the numbers with me").
Rules: evidence comes only from the transcript; where there is none, say so.
Transcript: [CALL TRANSCRIPT]
Use this when they admitted the pain and it now needs a number their CFO would respect.
AE
Discovery
ChatGPT
Help me turn a stated pain into defensible cost math. The pain, in the prospect's own words: [QUOTED PAIN]. What I actually know: [KNOWN FACTS — team size, salary bands, volumes, error rates, whatever I really have].
Build the cost model in three layers: 1) Direct cost — time or money burned, with the arithmetic shown step by step. 2) Indirect cost — downstream effects, described qualitatively unless I gave you numbers for them. 3) Risk cost — what a bad quarter of this looks like, qualitatively.
Rules: never invent a number. Every figure in the model must come from [KNOWN FACTS] or be an explicitly labeled assumption ("ASSUMPTION: fully loaded hourly cost — confirm with prospect"). Collect all assumptions in a block at the end, each phrased as a question I can validate on the next call. Present the final model as a short table plus a three-sentence plain-language summary a CFO would not roll their eyes at. If [KNOWN FACTS] is too thin for even the direct-cost layer, tell me exactly what to collect first.
Use this when the discovery call ended an hour ago and the recap should land before their memory fades.
AE
Discovery
ChatGPT
Write my post-discovery recap email to [PROSPECT NAME] from this transcript or my notes: [CALL TRANSCRIPT OR NOTES].
Structure: 1) A one-line thanks, no gushing. 2) "What I heard" — three bullets in the prospect's own vocabulary, quoting their phrases where possible, covering the problem, its impact, and what they have already tried. 3) "What we agreed" — next steps with owners and dates exactly as discussed; where a step had no date, propose one in brackets for them to confirm. 4) One clarifying question that proves I was listening, drawn from a moment where the conversation was ambiguous.
Rules: under 150 words. Zero product pitch — this email sells the accuracy of my listening and nothing else. Their words beat my words everywhere: where my notes conflict with their phrasing, use theirs. No "per my last", no corporate stiffness. Match the level of formality the transcript shows they use.
Use this when the deal feels good — and feelings are not a forecast category.
AE
Qualification
ChatGPT
Score this opportunity like a skeptical sales manager. Deal context: [DEAL NOTES — account, stage, size band, history]. Latest evidence: [CALL TRANSCRIPT OR SUMMARY].
Score seven dimensions from 0 to 3, each with cited evidence: identified pain, quantified impact, access to the economic buyer, a defined decision process, a timeline with a reason behind it, champion strength, and competitive position. 0 means no evidence; 3 means direct evidence in the prospect's own words. Every score cites a quote or a specific fact — never "seems positive".
Output: the scorecard table, a total with a verdict band (pursue hard / develop / qualify out), the two weakest dimensions with one concrete action each to strengthen them this week, and the single question I most need answered that nothing in my notes currently answers.
Rules: bias toward skepticism — where evidence is ambiguous, score low and say why. Do not soften the verdict to be nice. Under 400 words total.
Use this when the deal is warm but every thread runs through one person.
AE
Qualification
ChatGPT
Assess how dangerously single-threaded this deal is. Who I talk to now: [CURRENT CONTACTS AND TITLES]. What I know about the org: [ORG NOTES]. Deal size and stage: [DEAL SIZE AND STAGE].
Produce: 1) A thread map — each contact, their probable role in the decision, and the relationship strength my notes support (strong / warm / thin), one line each. 2) The gap list — roles a deal of this size typically requires that I have no relationship with: economic buyer, security or legal, procurement, the executive sponsor, the loudest skeptic. 3) For the two most dangerous gaps: a specific route in — who I already know who could introduce me, the meeting format that fits, and the exact words for the ask. 4) The single-threading risk stated bluntly: what happens to this deal if [PRIMARY CONTACT] changes jobs next month.
Rules: routes must go through people and context I actually listed — no fantasy referrals. Under 350 words.
Use this when discovery is done and a generic product tour would waste everything you learned.
AE
Discovery
ChatGPT
Design a demo agenda from my discovery evidence, not from our default product tour. Discovery evidence: [DISCOVERY NOTES OR TRANSCRIPT]. Attendees and roles: [ATTENDEE LIST]. What we sell: [YOUR PRODUCT].
Build: 1) A three-line opening that replays their problem in their words and states what this hour will prove — their success criteria, not our feature list. 2) An agenda of three or four chapters, each named after one of their stated problems ("Chapter: the Friday reconciliation mess"), with the one workflow to show, the one question to ask a named attendee mid-chapter, and what to deliberately skip. 3) A planned pause: where to stop and ask "how does this compare to how you handle it today". 4) A closing frame that converts interest into a concrete next step with an owner and a date.
Rules: everything shown must map to evidence in my notes — mark anything unsupported as CUT OR VALIDATE FIRST. Plan under 45 minutes of content for a 60-minute slot. Under 300 words.
Mid-funnel is where good discovery goes to die. Eight prompts for objections, stakeholders, stalls, and the political work of a real deal. If you live in this stretch of the funnel full-time, the ChatGPT prompts for account executives collection was built for your desk.
25
Objection Reframe Matrix
Use this when the same five objections keep landing and improvised answers keep producing varied results.
AE
Objections
ChatGPT
Build me an objection-handling matrix for selling [YOUR PRODUCT] to [PERSONA]. The five objections I hear most, in my prospects' actual words: [OBJECTION LIST].
For each objection, write three moves. 1) CLARIFY — a question that finds what is actually behind the words, since the stated objection is rarely the real one. 2) REFRAME — a two-to-three-sentence response that changes the comparison frame (cost versus cost of doing nothing, feature versus outcome, risk of acting versus risk of waiting) without dismissing their point. 3) ADVANCE — the single sentence that moves to a next step if the reframe lands.
Rules: spoken register — contractions, short sentences, words a rep would actually say on a live call, not paragraphs to read aloud. Never argue the prospect is wrong; validate the concern before reframing, every single time. No invented statistics and no invented customer references anywhere. Format as a clean matrix I can print. If any objection in my list is vague, ask me for the exact words prospects use before answering.
Use this when "it's too expensive" just landed and the next sixty seconds decide the deal's shape.
AE
Objections
ChatGPT
Coach me through price pushback on [YOUR PRODUCT], sold to [PERSONA]. Describe my list price structurally as [YOUR PRICING SHAPE — per seat, per month, one-time, tiers]. The pushback, verbatim: [EXACT WORDS THEY USED].
Write a realistic two-minute dialogue: their pushback, my response, their likely counter, my next response — three exchanges deep. My first response must diagnose which objection this actually is by asking one diagnostic question — no budget, budget committed elsewhere, anchored to a cheaper alternative, testing for a discount, or sticker shock with real interest — and my next responses must answer the diagnosed version, not the surface words.
After the dialogue, add the three concessions I could trade if needed, ordered from cheapest to me (a longer term, a case study, a reference call) to most expensive (price), each with the words for asking something in return — never a naked discount.
Rules: no defensiveness, no immediate discounting, no "you get what you pay for". Spoken register. Under 350 words.
Use this when the real competitor is "do nothing" and inertia is winning.
AE
Advancing
ChatGPT
The deal at [PROSPECT COMPANY] is stalled against the status quo, not against a competitor. Evidence of the pain they have admitted, quoted: [ADMITTED PAINS]. How long they have lived with it: [DURATION]. What doing nothing means operationally: [KNOWN CONSEQUENCES].
Write a one-page "cost of waiting" narrative addressed to [CHAMPION NAME], designed to be forwarded internally: 1) Open with their own admission, quoted. 2) Walk through the next two quarters if nothing changes — qualitative, concrete, grounded only in the consequences I listed. 3) Name the honest counterargument — changing tools has real cost too — and answer it with the smallest credible first step, not a big-bang pitch. 4) Close with one question that invites a decision about deciding: "is this worth solving this year — yes or no".
Rules: no invented numbers, no fear-mongering adjectives, no doom. Calm, specific, factual register — the reader should feel understood, not threatened. Under 250 words of output.
Use this when it is one deal, three audiences, and the CFO does not care what the end user loves.
AE
Advancing
ChatGPT · also strong on Claude
Take my core deal story and translate it for three stakeholders. The story: [DEAL STORY — problem, solution, evidence gathered during the evaluation so far]. The stakeholders: [STAKEHOLDER A — e.g. CFO], [STAKEHOLDER B — e.g. VP OF SALES], [STAKEHOLDER C — e.g. DAY-TO-DAY USER].
For each stakeholder, produce a five-sentence version: what they personally gain, what risk this reduces for them, what it costs them (money, change effort, political capital — be honest), the one proof point from my evaluation evidence that matters most to them, and the one question they will ask that I must not fumble — with a two-sentence answer.
Rules: same facts, three framings — the versions must never contradict each other, because these people will compare notes. Vocabulary must match each audience: financial terms for the CFO, operational terms for the VP, day-in-the-life terms for the user. No superlatives. Use only evidence contained in [DEAL STORY]; where I have no proof point for a stakeholder, flag the gap instead of inventing one.
Use this when verbal enthusiasm is high and calendar reality is nonexistent.
AE
Advancing
ChatGPT
Draft a mutual action plan working backward from [TARGET SIGN DATE] for the deal with [PROSPECT COMPANY]. Today's date: [TODAY'S DATE]. Current stage: [CURRENT STAGE]. Steps their side has mentioned: [KNOWN STEPS — security review, legal, procurement, exec approval, pilot].
Build a table with columns: milestone, owner (a name or role from my notes — never "TBD" without a flag), calendar week, and exit criterion (how both sides know it is done). Include the steps buyers forget to mention until they hurt: security questionnaire, data processing terms, procurement intake, signer availability. Insert buffer where slippage is likeliest and label it as buffer honestly.
After the table: 1) the two milestones most likely to slip, each with an early-warning sign; 2) a covering note to [CHAMPION NAME] proposing this plan as a draft for their edits — collaborative tone, their plan not mine, under 80 words.
Rules: dates must be arithmetically consistent with [TARGET SIGN DATE]; if the math does not fit, say the target is unrealistic and show the earliest credible date instead.
Use this when an active deal went silent two weeks ago and "just checking in" would bury it for good.
AE
Advancing
ChatGPT
An engaged deal has gone quiet. Last meaningful contact: [LAST CONTACT — what happened and when]. Where the deal stood: [STAGE AND CONTEXT]. My contact: [CONTACT NAME AND TITLE].
Give me three re-engagement options with genuinely different mechanics, plus the logic for choosing between them. Option 1 — the value-add: a message delivering one new, relevant thought (an insight about their industry, a question their team should be asking — derived from my context, not a gated PDF), with no ask attached. Option 2 — the process check: a blunt-but-kind note asking where this actually stands, making "it's dead" an easy answer ("if priorities moved, tell me and I'll close the file — no hard feelings"). Option 3 — the path-around: a note to a second contact from the evaluation, framed as continuity rather than going over anyone's head, plus the one-line courtesy note to my original contact.
Rules: each message under 90 words. Banned: "just checking in", "bubbling this up", "thoughts?". For each option, state the situation where it is the wrong move.
Use this when you know exactly who you are up against and product trash talk would backfire.
AE
Advancing
ChatGPT
We are head-to-head with [COMPETITOR] at [PROSPECT COMPANY]. My honest differentiators, as I understand them: [YOUR DIFFERENTIATORS]. Their honest strengths, as I understand them: [COMPETITOR STRENGTHS].
Build a question-led competitive strategy — questions that let the prospect discover the difference themselves. 1) Five evaluation questions to plant that map to my differentiators, phrased neutrally enough to sound like good diligence rather than sabotage ("how does each option handle X when Y happens"). 2) For each question: why it matters to the buyer on the merits — if you cannot defend it as genuinely good diligence, cut it. 3) The two questions THEY are probably planting against me, inferred from [COMPETITOR STRENGTHS], each with a two-sentence honest answer that concedes what is true and gives it context.
Rules: never write a claim about [COMPETITOR] beyond what I provided. No disparagement — the register is "help the buyer run a good evaluation". Spoken language for all suggested answers. Under 350 words.
Use this when the deal needs an exec-to-exec touch and your VP asks "what should I say?"
AE / Manager
Advancing
ChatGPT
Draft an email for MY executive, [MY EXEC NAME AND TITLE], to send to [PROSPECT EXEC NAME AND TITLE] at [PROSPECT COMPANY]. Deal context: [DEAL CONTEXT — stage, champion, what has been evaluated]. Why executive involvement now: [REASON — stalled approval, strategic account, they asked].
The email must sound like a busy executive wrote it: under 110 words, no sales vocabulary, peer register. Structure: one line of context ("our teams have been working together on..."), one substantive observation about the prospect's business that shows the exec is paying attention — drawn from [DEAL CONTEXT], not flattery — a clear statement of commitment ("you would have my attention on this"), and a soft offer of a short exec-to-exec call within a specific timeframe, positioned as optional.
Rules: my exec must not pitch features and must not undercut my process — "happy to discuss pricing" is banned. No urgency phrasing. Then add a two-line internal briefing note to my exec on why this send matters, so they actually send it.
Closing is a process, not an event. Eight prompts for proposals, negotiation, forecast honesty, and the operational hygiene that keeps next quarter alive while this one lands.
33
Executive Summary Assembler
Use this when the proposal needs a first page the economic buyer will actually read.
AE
Closing
ChatGPT
Write the executive summary page of my proposal for [PROSPECT COMPANY]. Inputs — their stated problem, in their words: [PROBLEM QUOTES]. What both teams validated during the evaluation: [EVAL FINDINGS]. Proposed scope: [SCOPE]. Their success metrics: [THEIR METRICS]. Commercial shape, described structurally: [COMMERCIAL SUMMARY].
Structure, in this order: the situation in two sentences using their vocabulary; what both teams confirmed during the evaluation (shared findings, not my claims); the proposed path stated as outcomes tied to [THEIR METRICS]; what we need from their side to succeed — honesty here builds trust, so name the effort their team must invest; and the decision being asked for, in one plain sentence.
Rules: one page maximum — under 280 words. Zero unverified claims: every statement must trace to [EVAL FINDINGS] or [PROBLEM QUOTES]. No superlatives about my company, no "industry-leading", no vision statements. Write it so the economic buyer who skipped every meeting understands the whole deal in ninety seconds.
Use this when procurement is on tomorrow's calendar and winging it is a strategy for losing margin.
AE
Closing
ChatGPT
Build my negotiation prep sheet for the final stretch with [PROSPECT COMPANY]. Deal shape: [DEAL SHAPE — product, term, seats or volume]. What they have signaled they will push on: [EXPECTED ASKS]. My constraints: [MY CONSTRAINTS — floor terms, approval thresholds, what I cannot give].
Produce: 1) A give-get table — every concession I might offer, its real cost to me (low / medium / high), and what to request in return: case study rights, a longer term, a faster signature, an introduction elsewhere in the org, expanded scope. Never give without getting. 2) My walk-away conditions, stated precisely, derived from [MY CONSTRAINTS]. 3) The three asks they will probably open with, each with a calm first response that neither concedes nor stonewalls. 4) Trades to propose if we deadlock, ordered by my preference.
Rules: no specific prices in your output — refer to terms structurally ("the annual rate", "the onboarding fee"). Spoken register for the response lines. One page, under 350 words.
Use this when the deal is "definitely closing this quarter" and you want that claim audited.
AE / Manager
Closing
ChatGPT
Stress-test my close plan like a deal-desk skeptic. The plan, however rough: [CLOSE PLAN]. Committed close date: [CLOSE DATE]. What the buyer has actually DONE so far — not said, done: [BUYER ACTIONS TO DATE].
Audit: 1) Every step that depends on buyer action with no evidence of buyer commitment — flag it HOPE, NOT PLAN. 2) The standard steps missing for a deal like this: security review, legal redlines, procurement intake, the signer's actual availability during the close week. 3) The date math walked backward from [CLOSE DATE] — is there physically enough calendar for the remaining steps, holidays included? Show the arithmetic. 4) A verdict — commit, best case, or pipeline — plus the single piece of buyer evidence that would upgrade it.
Rules: verbal enthusiasm counts as zero evidence; only actions count (booked meetings, sent documents, named signers). Be blunt — a plan that survives this audit should be one I would defend to a CRO. Under 300 words.
Use this when pipeline review is coming and every deal needs a story that survives questioning.
Manager
Pipeline ops
ChatGPT · also strong on Gemini
Act as my toughest sales manager and scrub my pipeline before the real review does. I will paste my open deals with stage, amount band, close date, next step, and last activity: [PIPELINE EXPORT].
For each deal, ask the four questions that expose weak forecasting — and answer them from my data where possible: When did the buyer last do something (not me)? Does the next step have a date and a buyer-side owner? What has changed since last month? Why does the close date sit in that month — buyer reason, or my hope?
Then classify each deal: COMMIT (evidence), BEST CASE (momentum with one gap), UPSIDE (early or stalled), or SCRUB CANDIDATE (no buyer activity in 30+ days) — and total the counts. For every scrub candidate, write the one honest question to ask the buyer this week that resolves it either way.
Rules: judge only from the data I pasted; where a field is missing, that absence IS the finding — say so. Output a table, then a five-line summary.
Use this when the deal closed and the implementation team deserves better than a CRM stub.
AE / CS
Post-close
ChatGPT
Write the handoff brief from sales to the onboarding and customer-success team for [PROSPECT COMPANY], closed on [CLOSE DATE]. Inputs — discovery and evaluation notes: [DEAL NOTES]. The people: [STAKEHOLDER LIST WITH ROLES]. What was promised: [COMMITMENTS MADE — scope, timelines, anything verbal].
Structure: 1) Why they bought, in their words — the pain that opened the deal and the metric they expect to move. 2) The people map: champion, economic buyer, skeptic to watch, day-to-day owner — one line each on how they engage. 3) Every commitment made during the sale, explicitly, including verbal ones — an implementation team surprised by a promise is how accounts sour. 4) Risks: where enthusiasm is thin, which team resisted, what nearly killed the deal. 5) The first 30 days: what success must look like by day 30 for the champion to look smart internally.
Rules: brutally honest — this document is internal, and optimism helps nobody. Under 400 words.
Use this when the deal just ended — either way — and the reason it ended is worth more than the outcome.
Manager
Post-close
ChatGPT
Build a win/loss interview kit for [DEAL NAME], which we [WON OR LOST] against [COMPETITOR OR STATUS QUO]. My internal theory of why: [YOUR THEORY].
Produce: 1) A ten-question interview guide for the buyer contact, sequenced from easy to candid: how the need emerged, who shaped the criteria, how the alternatives compared, where our process helped or hurt, what nearly changed the outcome, and what they would tell a peer who was evaluating us. Include the follow-up probe for each question ("what makes you say that"). 2) A neutral 60-word email requesting the 25-minute conversation — the tone must make honesty safe: no defensiveness for a loss, no victory lap for a win. 3) A one-page debrief template to fill in afterward: decision timeline, real criteria versus stated criteria, my theory confirmed or broken, and one process change to make next quarter.
Rules: every question must be outcome-neutral — it should work identically for a win or a loss. Never ask "why didn't you pick us" directly; the sequence gets there sideways.
Use this when it is your turn in the Monday meeting and the numbers need a narrative.
AE / Manager
Pipeline ops
ChatGPT
Turn my raw pipeline data into the three-minute narrative I will deliver at pipeline review. The data: [PIPELINE DATA — deals, stages, amounts banded, movement since last review]. My quota position: [QUOTA CONTEXT]. What changed since last review: [KEY CHANGES].
Structure the narrative: 1) The headline — my quarter in one sentence, stated plainly, no hedging. 2) The three deals that matter most and the specific buyer-side event each one needs next, with dates. 3) What moved since last review — in and out, with the honest reason for each slip; a slipped deal with a clear reason builds more credibility than a defended fantasy. 4) Where I need help — one specific ask of my manager: an exec touch, a resource, cover on a discount approval. 5) The risk I am watching that nobody has asked about yet.
Rules: numbers stay exactly as banded in my data, never invented or sharpened. First person, spoken register, 200 words maximum — it has to fit in three minutes with interruptions.
Use this when thirty minutes on Friday can save Monday's entire standup.
AE / Manager
Pipeline ops
ChatGPT
Run me through a 30-minute pipeline hygiene sprint. I will paste my open opportunities with whatever fields export easily: [PIPELINE EXPORT].
Generate my sprint checklist, ordered by damage prevented per minute: 1) Deals with close dates already in the past — list them, and for each give the one-line update-or-push decision. 2) Deals with no logged next step — draft the next step in seven words or fewer, based on stage and last activity. 3) Deals with no buyer contact in 21+ days — flag each with the right re-engagement route based on my last-activity note: value-add, process check, or path-around. 4) Stage inflation suspects — deals whose stage implies evidence my data does not show; name the missing evidence for each. 5) The two records to fix first if I only get ten minutes.
Rules: work only from my pasted data; a missing field is a finding, not a blank to fill creatively. Output as a numbered checklist I can work top to bottom, time-boxed to 30 minutes total.
These prompts are built to chew on source documents — transcripts, job postings, pipeline exports, research notes. Most of them refuse to invent what you did not paste. That refusal is the feature: output quality tracks input honesty, so bring the raw material.
Fill every bracket before you run it
An ALL-CAPS bracket left unfilled is a guess you just authorized. [PROSPECT COMPANY], [CALL TRANSCRIPT], [YOUR DIFFERENTIATORS] — each one is a fuel line. Thirty seconds of filling brackets beats three rounds of correcting generic output.
Work one stage per week
Forty prompts is a library, not a to-do list. Pick the stage your quarter actually hinges on — cold outreach in a dry month, forecast scrubs before review season — and run those eight until they are muscle memory. Then move down the funnel.
Keep the constraints in
The word caps, banned phrases, and ask-me-first lines are the anti-slop machinery. Deleting them feels harmless and costs you everything — they are the difference between an email a human wrote and an email a template produced.
Make the last pass yours
Every draft leaves these prompts at 90 percent. Rewrite one line in your own voice, cut anything you would not say out loud, and send it as you. The rep is the seller; the prompt is the instrument.
Questions
ChatGPT prompts for sales: questions, answered
What are the best ChatGPT prompts for sales?
The best ChatGPT prompts for sales are specific to a stage and a job: researching an account, writing a first touch, running discovery, handling an objection, or scrubbing a forecast. The 40 prompts on this page cover all five stages, each with defined inputs, an output format, and constraints that keep the answer from going generic. Start with the stage your week actually depends on.
Do these prompts work in Claude and Gemini too?
Yes. Every prompt here is written in plain instruction language, so it runs in ChatGPT, Claude, or Gemini without edits. A few are tagged where we found them notably strong in one tool — Claude tends to handle long call transcripts well, and Gemini is comfortable with pasted spreadsheet-style pipeline data — but those are working observations, not benchmarks.
Are these ChatGPT prompts for sales free to use?
All 40 prompts on this page are free, ungated, and fully copyable — no email required. They are a sample of the Promptifi library, which holds 2,900+ tested B2B sales prompts. Pro access is $4.99/mo, $48/yr, or $99.00 lifetime, with lifetime capped at 500 seats.
How do I fill in the bracketed variables?
Every ALL-CAPS bracket like [PROSPECT COMPANY] or [CALL TRANSCRIPT] marks an input the prompt needs from you. Replace each one with your real detail before you run it — paste the actual transcript, the actual job posting, the actual pipeline export. The more real material you feed a prompt, the less generic its output can possibly be.
Will a prospect know my email was written with ChatGPT?
Only if you send the first draft. These prompts are built to fight the tells — word caps, banned filler phrases, and instructions to use the prospect's own vocabulary — but the final pass is yours. Rewrite one line in your own voice, cut anything you would not say out loud, and the result reads like you on a good day.
40 of 2,900+
The count stops at 40. The library keeps going.
Every prompt in the Promptifi library — 2,900+ of them — is tested, tagged by role, stage, and tool, and written to the same standard as the forty above. When the free shelf runs out mid-quarter, the building is open.