Updated weekly · 10 prompts

AI Sales Prompts

Cold outreach that gets replies, discovery calls that uncover real pain, objection handling that doesn't sound canned — tested across ChatGPT, Claude, and Gemini.

AI sales prompts are structured instructions you hand to a tool like ChatGPT, Claude, or Gemini so it can help you do the unglamorous parts of selling — writing a cold email that doesn't get deleted in two seconds, building a discovery agenda that uncovers real budget, scripting a response to "we'll think about it," or drafting a proposal at 9pm before tomorrow's call. They are not replacements for talking to humans. They are leverage on the prep, the writing, and the synthesis.

This page is built for SDRs, AEs, account managers, and founders selling their own product. If you spend half your week typing the same kinds of emails, recapping the same kinds of calls, and answering the same kinds of objections, these prompts compress that work without making you sound like a bot. Each one is written like an experienced sales coach would write it: clear role, specific deal context, and an output format you can actually paste into Salesforce, HubSpot, or your inbox.

The real unlock is specificity about the deal. A vague prompt ("write me a cold email") returns the same generic copy every other rep sends. A specific prompt ("write three cold emails to a VP of Operations at a 200-person logistics company who just hired a new CFO, referencing a recent layoff round, in under 90 words each") returns something a buyer might actually open.

Why these prompts work

Every prompt on this page follows the same recipe: Role + Deal Context + Output Format. The role tells the AI who to be ("act as a senior enterprise AE who has closed seven-figure deals"). The deal context grounds it in real specifics — buyer title, company size, prior touchpoints, known objections. The format pins down what good looks like — a 90-word email, a 30-minute discovery agenda, a one-page proposal. When all three are present, the AI stops sounding like every other sales tool and starts sounding like a thoughtful rep who actually knows the account.

Free Prompts You Can Copy Today

Six free prompts followed by four advanced ones from the Sales Pack. All six free prompts include a short "Why this works" explainer so you can adapt them to your own pipeline.

1

Cold Email Generator (3 versions)

Sales
You are a senior B2B sales writer who has shipped thousands of cold emails that actually get replies. Write three different cold email versions to the prospect described below. What we sell: [PRODUCT OR SERVICE IN ONE SENTENCE] Who I'm emailing: [NAME, TITLE, COMPANY, COMPANY SIZE] Why this person, right now: [TRIGGER EVENT — FUNDING, NEW HIRE, LAYOFFS, PRODUCT LAUNCH, EARNINGS, ETC.] The pain we solve for them: [SPECIFIC PAIN — NOT GENERIC] Proof we can mention: [ONE CUSTOMER LOGO, METRIC, OR CASE STUDY] Call to action: [15-MIN INTRO CALL / SHORT DEMO / OTHER] Write three versions, each under 90 words: Version A — Direct: Lead with the trigger event and a sharp value claim. Version B — Curious: Open with a real, specific question instead of a pitch. Version C — Proof-led: Open with the customer outcome and let it carry the email. Rules: no "I hope this finds you well." No "quick question." No paragraphs longer than two sentences. Subject lines under 6 words. End each with one clear, low-friction CTA.
Why this works
Asking for three distinct angles — direct, curious, proof-led — forces the AI to stop hedging and commit to a stance per email. The banned phrases ("I hope this finds you well," "quick question") quietly strip out the openers buyers ignore on sight. The 90-word ceiling keeps you in inbox-preview territory.
2

Discovery Call Agenda

Sales
Act as a senior enterprise AE who runs disciplined discovery calls. Build a structured agenda for a [LENGTH IN MINUTES]-minute discovery call. Deal context: - What we sell: [PRODUCT] - Buyer's title and team: [TITLE / FUNCTION] - Company: [NAME, SIZE, INDUSTRY] - How the meeting was booked: [INBOUND / OUTBOUND / REFERRAL / EVENT] - What we know already: [2-3 FACTS — TECH STACK, RECENT NEWS, PAIN HINTS] - The decision we want by end of call: [NEXT STEP — DEMO, TECHNICAL CALL, PROPOSAL] Output the agenda as a table with these columns: Minute range, Section, Goal of this section, 2-3 questions to ask, What "good" looks like. Cover at minimum: rapport opener, agenda set, current state, pain and impact, decision process and timeline, budget signals, recap, and next step. End with three follow-up questions I should send by email if anything was unclear. Keep questions open-ended. No leading questions. No "yes/no" traps.
Why this works
Forcing the agenda into a table with minute ranges turns vague "discovery best practices" into a concrete plan you can run. The "what good looks like" column is the trick — it makes the AI tell you the signal you're listening for, not just the question you're asking.
3

Objection Handling Script

Sales
You are a sales coach who has trained AEs through hundreds of late-stage objections. Build me a clean objection handling script for the situation below. Deal context: - What we sell: [PRODUCT] - The buyer's role: [TITLE] - The deal stage: [DISCOVERY / EVAL / NEGOTIATION / RENEWAL] - The exact objection (verbatim if possible): [PASTE THE OBJECTION] - What we suspect is really going on underneath it: [BUDGET FREEZE / COMPETITOR / FEAR / TIMING / OTHER] Produce, in this order: 1. The reframe in one sentence — what the objection probably actually means 2. A short acknowledgement script (no more than 25 words) that doesn't fight the buyer 3. One diagnostic question that surfaces the real blocker 4. Two ways to respond if the answer is "budget" vs. "timing" vs. "competitor" 5. A concrete next-step ask that moves the deal forward without being pushy 6. A red-flag list: 3 signs this objection is actually a polite "no" Tone: calm, curious, never defensive. Don't use the word "absolutely." Don't promise anything you'd need legal to approve.
Why this works
The "reframe" step asks the AI to translate the surface objection into the real one — which is where most reps lose deals. The red-flag list forces the model to be honest about when an objection is actually a soft no, instead of pretending every deal is winnable.
4

Sales Proposal Writer

Sales
Act as a senior account executive who writes proposals that close. Draft a one-page proposal for the deal below. Deal context: - Buyer: [NAME, TITLE, COMPANY] - What they told us they want to solve: [PAIN IN THEIR WORDS] - The metric they care about: [REVENUE, RETENTION, COSTS, TIME, RISK] - Stakeholders involved: [WHO ELSE NEEDS TO REVIEW] - Timeline they mentioned: [WHEN THEY WANT TO START] - Pricing: [LIST PRICE, TARGET DISCOUNT, CONTRACT TERM] - Two competitors they're also evaluating: [LIST] Structure the proposal as: 1. Executive summary (3 sentences max — recap their goal in their language) 2. Recommended solution (which package, why it fits) 3. Expected outcomes (concrete, with rough metric ranges, no overpromising) 4. Implementation plan (week-by-week for the first 30 days) 5. Investment (table with line items, term, total, and any one-time fees) 6. Why us, in one short paragraph (no logo soup) 7. Next step with a date Voice: confident, specific, no jargon. Avoid "synergy," "unlock," "leverage." Write so a CFO could skim it in 90 seconds and say yes.
Why this works
Pinning the executive summary to "recap their goal in their language" is the move — proposals lose when reps describe the problem in vendor-speak instead of buyer-speak. The "skimmed in 90 seconds by a CFO" test forces tightness most AI proposals never get to.
5

ICP Definition (Ideal Customer Profile)

Sales
You are a B2B GTM strategist who builds ICPs that sales and marketing actually use. Help me define a sharp Ideal Customer Profile. Inputs I'll give you: - What we sell: [PRODUCT, IN ONE SENTENCE] - Our 3-5 best current customers: [NAMES, INDUSTRIES, SIZES, USE CASE, ANNUAL CONTRACT VALUE] - Our 2-3 worst-fit customers (churned, painful, low expansion): [NAMES, WHY THEY WERE BAD FITS] - Markets we explicitly do not want to chase: [LIST] Produce: 1. A one-paragraph ICP definition in plain English 2. A firmographic table: industry, company size, geography, revenue band, tech stack signals, funding stage 3. A persona table: title, team, top 3 pains, top 3 goals, where they hang out online, how they buy 4. Three "yes" qualifiers — if you see these, lean in 5. Three "no" disqualifiers — if you see these, walk away 6. A one-sentence "anti-ICP" describing who this is NOT for Be ruthless. A good ICP excludes more than it includes. If our best customers contradict our worst, say so explicitly.
Why this works
Most ICP exercises are too inclusive to be useful. Forcing both "yes qualifiers" and "no disqualifiers" — plus an explicit anti-ICP — turns the output into a decision tool reps can actually run a list against, not a marketing poster.
6

Lead Scoring Framework

Sales
Act as a RevOps lead who has built lead scoring models for fast-growing B2B teams. Help me build a simple, defensible lead scoring framework. Context: - What we sell: [PRODUCT] - Average sales cycle: [DAYS OR MONTHS] - Average contract value: [DOLLARS] - Lead sources we currently get: [INBOUND DEMO REQUEST, NEWSLETTER, WEBINAR, OUTBOUND, REFERRAL, ETC.] - Tools we use: [HUBSPOT / SALESFORCE / OTHER] - Our ICP, in one line: [SUMMARY] Output: 1. A scoring model with two dimensions: Fit (firmographic + persona match) and Intent (behavioral + engagement signals) 2. For each dimension, a table of 6-8 attributes with point values, max points, and how to capture the data 3. A simple thresholds map: Cold (0-20), Warm (21-50), Hot (51-80), MQL-to-SDR (81+) — with what each tier should trigger 4. Three signals worth more than they look (sleeper indicators) 5. Three vanity signals to ignore even though they feel important 6. A 30-day plan for tuning the model with real win/loss data Keep it simple enough that an SDR can apply it without a data team.
Why this works
Splitting the model into Fit and Intent is the standard pro move — it stops the AI from collapsing everything into a single confused score. The "sleeper indicators" and "vanity signals" sections force the model to make real calls about what predicts revenue, not just what's easy to track.
7

Sales Call Summary

Sales
You are a senior AE writing a clean post-call summary for both internal CRM notes and the prospect's inbox. I'll paste the call transcript or my raw notes below. Deal context: - Buyer: [NAME, TITLE, COMPANY] - Stage: [DISCOVERY / DEMO / EVAL / NEGOTIATION] - Date of call: [DATE] - Who else was on the call: [NAMES AND ROLES] Produce two artifacts: 1. CRM-facing summary — pain points heard, decision criteria, stakeholders, timeline, budget signals, competitors mentioned, MEDDPICC fields, risk score 1-5 with reasoning 2. Buyer-facing recap email — short, friendly, mirrors their language, lists agreed next steps with dates, ends with one clear ask...
8

Win/Loss Analysis

Sales
Act as a sales effectiveness consultant. I'll paste a batch of recent closed-won and closed-lost deals below — names, ACVs, competitors, stages where they died, and any notes from the AE. Produce a structured win/loss analysis. Output: patterns across wins, patterns across losses, the most expensive single reason we lose, the cheapest improvement we could make next quarter, three coaching recommendations by rep persona, and a one-page summary I can show the leadership team...
9

Renewal Pitch Builder

Sales
You are a senior account manager who runs renewals at a 95%+ gross retention SaaS. Build me a renewal pitch for the account below. Account context: customer name, contract end date, ACV, usage trends, key wins shipped this year, support tickets, executive sponsor changes, expansion opportunities, and any churn risk signals. Produce: a renewal narrative anchored on outcomes already delivered, three expansion plays ranked by likelihood, a price defense if they push back, a contingency plan if the sponsor is gone, and a clean meeting agenda...
10

Sales Email Follow-Up Sequence

Sales
Act as a top-performing AE who actually closes deals from cold sequences. Build me a 6-touch follow-up sequence over 21 days for a prospect who didn't reply to the first cold email. Inputs: ICP, prospect title, what we sell, the original cold email, and the trigger event we used. Produce six emails — each with a subject line, body under 90 words, and a clear single CTA. Mix angles across the six: value-add, social proof, soft break-up, peer reference, content drop, hard break-up...

How to use these prompts well

Reps who get great output from AI tools follow the same loop, regardless of which model they use. Read the prompt before pasting it. Replace every bracketed placeholder with real, specific deal context — not "a software company" but "a 240-person fintech that just raised a Series B and uses Salesforce, Outreach, and Gong." The more concrete the inputs, the more concrete the output.

Then, edit the response. Treat the AI as a fast first-draft writer, not a final author. Cut anything that sounds like every other sales email you've ever received. Replace generic claims with the specific number or customer story you actually have. If the model produces something that contradicts what you know about the buyer, say so out loud — "the buyer is technical, not commercial, rewrite assuming they care about implementation risk over ROI" — and watch the second draft snap into focus.

Finally, save the prompts that work for your motion. The best reps build a small, personal library of three or four variants — one for cold outreach, one for discovery prep, one for proposal writing, one for objection responses — and refine them deal by deal. The prompt is a tool, not a script, and your edge is in how sharply you tune it for your buyer.

Related learning

Short, plain-English reads that pair well with this category.

Sales · 8 min read

ChatGPT for Sales: A Practical Playbook for Reps and Founders

How working AEs use ChatGPT for prospecting, call prep, and follow-ups — with real prompt patterns and the mistakes that flatten reply rates.

Read post →
Proposals · 7 min read

AI Prompts for Freelance Proposals That Actually Win

A walkthrough for solo operators and consultants — turning a discovery call into a tight, priced proposal without sounding like a template.

Read post →
Business · 8 min read

How to Use AI Prompts for Business (Without Wasting Your Team's Time)

A practical playbook for picking the right prompt, customizing it for your context, and editing the output so it lands in the room.

Read post →

Want the full Sales Pack?

The Sales Pack on TopAIPrompts bundles 50+ advanced sales prompts — multi-touch outbound sequences, MEDDPICC qualification scripts, executive briefing builders, renewal playbooks, win/loss synthesizers, deal-review frameworks, and more. One-time payment, lifetime updates.

  • 50+ sales prompts
  • CRM-ready templates
  • No subscription
Browse the Sales Pack

One new prompt every Tuesday.

Free. No spam. Unsubscribe anytime.

Prompt copied to clipboard