ChatGPT Prompts for Sales: 40 Prompts for Prospecting 2026
By Amy Johnson · November 14, 2026 · 17 min read
Key Takeaway
ChatGPT prompts for sales work when you feed them real context: your offer, the buyer's role, one verified fact about the account, a word limit and one goal. Save the context block, use the 40 prompts below for research, cold email, follow-ups, objections, discovery prep and LinkedIn, and source prospect facts from live data instead of the model's memory.
ChatGPT prompts for sales are reusable instructions that tell ChatGPT who you sell to, what you sell and what you want back, so it can draft research summaries, cold emails, follow-ups and call prep in seconds. The quality of the output depends almost entirely on the context you put in.
This guide gives you 40 copy-ready prompts across six parts of the sales cycle. Each one has placeholders in [brackets] and a short note on when to use it, plus a reusable context block that makes every prompt sharper.
TL;DR
- Context beats clever wording: give ChatGPT a role, your offer, the buyer, one verified fact, a word limit and one goal. Generic input gets generic emails.
- 40 prompts, six jobs: prospect research (1–6), cold email (7–15), follow-ups (16–21), objection handling (22–27), discovery prep (28–34) and LinkedIn messages (35–40).
- ChatGPT is a writer, not a database: it cannot verify emails or know who changed jobs last month. Pull those facts from an enrichment tool, then paste them into the prompt.
How to Use ChatGPT for Sales: Context Is Everything
The fastest way to use ChatGPT for sales is to treat it like a new SDR on day one. It writes fast, but it knows nothing about your product, your buyer or this specific account until you tell it.
That matters because selling time is scarce. Salesforce research found reps spend just 28% of their week actually selling, with the rest going to tasks like deal management and data entry. Good prompts give some of that time back.
OpenAI’s own prompt engineering guide recommends the same few habits for any task. Give clear instructions, supply the data the model would not otherwise have, show examples, and separate sections with headers or tags.
For sales, that turns into six ingredients in every prompt:
| Ingredient | Weak version | Strong version |
|---|---|---|
| Role | (none) | “You are an SDR who writes short, plain cold emails” |
| Offer | “our software” | One sentence on what it does and the outcome |
| Buyer | “a prospect” | Job title, company size, industry |
| Account fact | “personalize it” | A verified signal, such as a job posting or tech stack detail |
| Constraints | “keep it short” | “Under 90 words, no exclamation marks” |
| Goal | “get a meeting” | “Ask if a 15-minute look is worth it” |
One more rule: always edit before you send. As Zapier’s guide to ChatGPT for sales warns, a confidently wrong detail in a cold email is worse than no detail at all.
The Reusable Context Block
Paste this block at the top of a new chat, or save it in ChatGPT’s custom instructions or a project. Fill it in once, and every prompt below gets better without extra typing.
About me: I am a [role] at [company].
What we sell: [one sentence, plain language]
Who buys it: [job titles] at [industry] companies with [size] employees in [regions]
Problems we solve: [2 or 3 problems, in the buyer's words]
Proof: [1 or 2 real customer results with real numbers]
Differentiator: [the one thing competitors cannot say]
Price range: [range]
Tone: plain, direct, friendly. Short sentences. No buzzwords, no exclamation marks.
Never: invent statistics, customer names or facts about a prospect. If you need a fact you do not have, ask me.The last line does the most work. It stops ChatGPT from inventing customer results or prospect facts, which is the most common way AI sales copy goes wrong.
If you write with Claude as well, the same block works there. Our Claude cold email prompts and the free SyncGTM sales prompts library use the same structure.
Prospect Research Prompts (1–6)
Research prompts turn raw text into a short brief you can act on. Always paste the source text in; do not ask ChatGPT what it “knows” about a company, because its answer may be out of date or wrong.
1. Account snapshot from the company website
Act as a B2B sales researcher. Below is text from [Company]'s homepage, about page and pricing page.
Summarize in under 120 words:
- What they sell and who buys it
- Their business model (self-serve, sales-led, or both)
- Likely size and stage
- One thing that changed recently, if the text shows it
Only use facts from the text. Write "unknown" if the text does not say.
<company_text>
[paste text]
</company_text>When to use it: The "only use facts from the text" line is the most important one. Without it, ChatGPT fills gaps with plausible guesses.
2. Pain points by role
I sell [product] to [industry] companies with [size] employees. My buyer is the [job title].
List the 5 problems this person most likely owns that my product affects. For each one, give:
1. The problem in their words (how they would say it in a meeting)
2. What it costs them (time, money, risk or headcount)
3. One question I could ask to confirm it
Rank them from most to least urgent for a company at [stage].When to use it: Use this before writing a new sequence. The "in their words" output gives you subject lines and openers that sound like the buyer, not like your website.
3. Read a hiring signal
Here are 3 job postings from [Company]. I sell [product], which helps with [outcome].
Tell me:
- What team they are building and why, based on the postings
- Which posting is most relevant to my product, and the exact line that shows it
- A one-sentence reason to reach out now that references the posting
<job_postings>
[paste postings]
</job_postings>When to use it: Hiring is one of the clearest buying signals. A company hiring three SDRs has an outbound problem this quarter, not someday.
4. Tech stack angle
[Company] uses these tools: [list from tech stack data].
My product [product] integrates with [tools] and replaces [tools].
1. Which tools in their stack make my product easier to adopt?
2. Which tools suggest a gap my product fills?
3. Write one opening line for a cold email that references their stack without sounding like I spied on them.When to use it: Tech stack data turns "we integrate with your CRM" into a specific, checkable reason to talk.
5. Map the buying committee
I am selling [product] at a typical price of [price] per year to a [size]-employee [industry] company.
Map the likely buying committee:
- Economic buyer
- Champion
- Day-to-day user
- Technical evaluator
- Likely blocker
For each role, give the common job titles, what they care about, and the one message that would make them say yes to a first call.When to use it: Use it to decide who to multithread to after your first contact replies.
6. ICP fit check for a list
My ideal customer profile: [industry], [employee range], [region], uses [tool], and shows [signal].
Score each company below from 1 to 5 for fit. Give a one-line reason for each score. Mark any company where the data is missing as "needs research" instead of guessing.
<companies>
[paste rows: name, industry, employees, region, tools, recent signal]
</companies>When to use it: Good for small lists of 20 to 50 rows. For thousands of rows, run the scoring inside your enrichment table instead of a chat window.
For a deeper look at finding accounts with AI, see our guide to lead generation with ChatGPT.
Cold Email Prompts (7–15)
These ChatGPT prompts for sales emails all follow one pattern: a verified fact first, one problem, one outcome and one easy question. Start with prompt 7, then use the others to tighten, test and vary it.
7. The master cold email prompt
You are an SDR who writes short, plain cold emails that get replies.
Write a cold email to [first name], [job title] at [company].
What I sell: [one sentence]
Who it helps: [ICP]
Proof: [one customer result I can stand behind]
Fact about this prospect: [verified signal, e.g. hiring 3 SDRs, just launched X]
Rules:
- Under 90 words
- Open with the fact, not with me
- One problem, one outcome, one question
- Call to action: ask if it is worth a 15-minute look, no calendar link
- No "I hope this finds you well", no "I came across your profile", no exclamation marksWhen to use it: This is the prompt to save first. Every other cold email prompt below is a variation on it.
8. Cut it to the bone
Rewrite this email to under 70 words. Keep the opening fact and the call to action. Remove every sentence the reader could skip without losing meaning. Use words a busy person can read on a phone in 10 seconds.
[paste email]When to use it: ChatGPT drafts run long. A second pass for length almost always improves the email.
9. Signal-based opener
Write 5 different opening lines for a cold email to [job title] at [company].
The signal: [e.g. they posted a job for a RevOps manager on (date)].
Each line must:
- Mention the signal in a natural way
- Connect it to a problem my product solves: [problem]
- Be under 25 words
- Not start with "I"When to use it: Pick one line, then plug it into prompt 7. The signal does the personalization work, so the rest of the email can stay simple.
10. Subject lines
Give me 10 subject lines for this cold email. Rules: 2 to 5 words, lowercase, no clickbait, no emojis, nothing that looks like a newsletter. Mix three styles: the prospect's company name, the problem, and a question.
[paste email]When to use it: Short internal-looking subject lines tend to outperform clever ones. Test two at a time.
11. Problem-first email for a persona
Write a cold email to a [job title] at a [size]-employee [industry] company. Lead with the problem: [problem]. Describe what that problem looks like on a normal Tuesday for them in one sentence. Then give one line on how [product] fixes it, and close with a yes-or-no question. Under 80 words.When to use it: Use this when you have no account-specific signal. A sharp, familiar problem is the next best thing.
12. Case-study email (no invented proof)
Write a cold email that uses this customer story as proof:
Customer: [name or "a 40-person agency"]
Before: [situation]
After: [result, with the real number]
Use only the numbers I gave you. Do not add any statistic, percentage or customer name that is not in this prompt. Under 90 words, end with a question.When to use it: ChatGPT will happily invent a "37% lift" if you let it. The explicit ban on new numbers stops that.
13. First lines for a batch of prospects
For each prospect below, write one personalized first line for a cold email, under 20 words. Base it only on the "note" column. If the note is empty or vague, write "SKIP" instead of making something up.
Return a table: name | company | first line
<prospects>
[paste: name, company, title, note]
</prospects>When to use it: The SKIP rule keeps weak rows out of your sequence. Feed it real notes, such as a recent post or a job posting, not just a job title.
14. Get it critiqued by a skeptical buyer
Act as a busy [job title] who gets 40 cold emails a week and replies to almost none.
Read this email and tell me honestly:
1. Would you open it? Why or why not?
2. The exact line where you would stop reading
3. What would make you reply
4. A rewrite that fixes the top problem
[paste email]When to use it: Role-play critique is one of ChatGPT's best sales uses. It finds the weak line faster than a manager review.
15. One email, three personas
Here is a cold email written for a [job title A]. Rewrite it twice more: once for a [job title B] and once for a [job title C] at the same company. Change the problem and the outcome to match what each role is measured on. Keep each version under 90 words.
[paste email]When to use it: Useful when you multithread into one account. Each person gets a message about their own goals.
Before you load any draft into a sequence, run it through our cold email checklist. For more on AI personalization, read how to use AI to personalize cold emails.
Follow-Up Email Prompts (16–21)
Most replies come after the first email, so follow-ups deserve the same care. Each prompt below asks ChatGPT to add something new instead of repeating the first message.
16. Follow-up that adds something new
I sent this cold email 4 days ago and got no reply:
[paste email]
Write a follow-up under 60 words that adds one new thing: [a resource, a stat with its source, a customer example, or a question]. Do not say "just following up", "bumping this" or "circling back". Reply in the same thread, so no subject line.When to use it: Each follow-up should give the prospect a new reason to reply, not a reminder that you exist.
17. Plan a 4-step sequence
Plan a 4-email cold sequence to [job title] at [ICP]. For each step give: day to send, the angle (problem, proof, resource or breakup), a subject line if it starts a new thread, and the full email under 80 words. Each email must work on its own if it is the only one they read.When to use it: Ask for the angle per step first. It forces variety, so step 3 is not a reworded step 1.
18. Post-demo recap from call notes
Turn these demo notes into a recap email to [name]. Include:
- Their goals in their own words
- The 2 or 3 parts of the demo that mattered most to them
- Open questions and who owns each one
- The agreed next step with a date
Under 180 words. Friendly and direct.
<notes>
[paste notes or transcript]
</notes>When to use it: Send it within the hour. A recap in their words is also the email your champion will forward internally.
19. Re-engage a ghosted deal
A prospect went quiet after [last stage, e.g. the pricing call] 3 weeks ago. Context: [what they wanted, what was agreed].
Write a short email that:
- Does not guilt them
- Offers an easy way out ("if priorities changed, just say so")
- Gives one useful thing related to their goal
Under 70 words.When to use it: Giving people permission to say no often gets more replies than pushing for a yes.
20. Breakup email
Write a final email in a cold sequence to [job title]. Keep it under 50 words. Say this is the last email, restate the problem in one line, and ask one question they can answer in a single word. No passive-aggressive tone.When to use it: Breakup emails often get replies because they lower the pressure. Keep them polite so the door stays open.
21. Email your champion can forward
My champion [name, title] likes our product but needs approval from [boss's title]. Write an email my champion can forward to their boss. It should explain in under 150 words: the problem, the cost of doing nothing, what [product] does, the price, and the next step. Write it from the champion's point of view, not mine.When to use it: Champions rarely sell as well as you do. Hand them the exact words.
Need more examples to feed in as style references? Our follow-up emails for prospects who go silent make good few-shot examples for prompts 16 and 19.
Objection Handling Prompts (22–27)
ChatGPT is very good at playing a difficult buyer. Use these prompts to prepare responses before calls, then practice them out loud.
22. Build an objection library
I sell [product] at [price] to [ICP]. List the 10 most common objections this buyer will raise. For each one, give:
1. What they say
2. What they usually mean
3. A 2-sentence response that asks a question back
Group them under: price, timing, competitor, authority and need.When to use it: Review the list with your best rep and keep the responses that sound like a human said them.
23. "We already use [competitor]"
A prospect says: "We already use [competitor]."
Write 3 responses I could give on a call. Do not trash the competitor. Each response should acknowledge their choice, ask what they would change about it, and mention one specific difference: [your real differentiator]. Under 40 words each.When to use it: The goal is a conversation about gaps, not a feature war.
24. "It's too expensive"
A [job title] says our price of [price] is too high. Their current process costs them [time or money, if known].
Give me:
- 2 questions to find out if this is about budget, value or timing
- A short response that ties price to the cost of their current process
- One option to reduce risk (pilot, smaller scope, or monthly terms) that I can actually offer: [options]When to use it: List the concessions you are allowed to make. Otherwise ChatGPT will offer discounts you cannot give.
25. "Just send me some info"
On a cold call, the prospect says "just send me some info." Write 3 replies I can say that politely ask for 60 seconds to make sure what I send is relevant. Each reply should end with one qualifying question about [the problem my product solves].When to use it: "Send me info" is usually a polite exit. A good question tells you if there is a real deal behind it.
26. Objection role-play
Role-play as a skeptical [job title] at a [size]-employee [industry] company. I am selling [product]. Push back with realistic objections, one at a time. After each of my answers, rate it from 1 to 10 and tell me what a stronger answer would include. Do not go easy on me. Start now with your first objection.When to use it: Ten minutes of this before a big call beats reading a battlecard.
27. "Not a priority right now"
A prospect says "this isn't a priority right now." Write a reply email under 70 words that respects that, asks what is the priority this quarter, and asks if it is OK to check in at [specific month]. Mention one trigger that would make it urgent for them: [e.g. a new hire, a failed tool, a quarter-end target].When to use it: Log the month they give you and set a task. Timing objections are leads for next quarter.
For the frameworks behind these responses, read how to overcome objections in B2B sales.
Discovery Call Prep Prompts (28–34)
Discovery prep is where research pays off. These prompts build on the brief from prompt 1, so run that first and paste the output in.
28. Discovery questions from your research
I have a discovery call with [name, title] at [company]. Here is what I know: [paste research summary from prompt 1].
Write 10 open-ended discovery questions in the order I should ask them. Start broad (goals, current process), then go narrow (pain, cost, timeline, decision process). Mark the 3 questions I must not skip.When to use it: Feed it real research, not just a company name, or you get the same 10 questions every rep asks.
29. One-page pre-call brief
Build a one-page call brief from this information. Sections: company in 3 lines, the person and their likely goals, recent signals, 3 hypotheses about their pain, what success looks like for this call, and 2 risks to watch for. Bullet points only.
<info>
[paste website notes, LinkedIn summary, job posts, CRM notes]
</info>When to use it: Paste it at the top of your call notes so you can glance at it while you talk.
30. Hypotheses to test
Based on this account research, give me 3 hypotheses about why [company] might need [product] now. For each one, write the question that would prove it right or wrong, and what I should say if it is wrong.
[paste research]When to use it: Going in with hypotheses keeps discovery focused and shows the buyer you did homework.
31. MEDDPICC gap check
Here are my notes from 2 calls with [company]. Map them to MEDDPICC (Metrics, Economic buyer, Decision criteria, Decision process, Paper process, Identify pain, Champion, Competition). For each letter, say what I know, what is missing, and the exact question to ask next.
<notes>
[paste notes]
</notes>When to use it: Swap in BANT or SPICED if that is what your team uses. The structure is what matters.
32. Agenda email before the call
Write a short email to [name] for our call on [date]. Confirm the time, share a 3-point agenda based on [their stated goal], and ask if there is anything else they want to cover or anyone else who should join. Under 80 words.When to use it: The "anyone else who should join" line often brings the real decision-maker into call one.
33. Transcript to CRM notes
Summarize this call transcript into CRM notes. Format:
- Pain (in their words)
- Current tools and process
- Budget and timeline signals
- Decision-makers mentioned
- Objections raised
- Agreed next step and date
Quote the prospect directly where possible. Do not add anything that is not in the transcript.
<transcript>
[paste transcript]
</transcript>When to use it: Check your company's policy before pasting call recordings into any AI tool.
34. Mock discovery call
Play [name], a [job title] at [company], in a discovery call. Use this background: [research]. Answer my questions like a real buyer: be busy, a little guarded, and only share pain if I ask good follow-up questions. After 8 exchanges, stop and give me feedback on what I missed.When to use it: Practice the follow-up question, not just the first question. That is where deals get found.
Still trying to land the call? Pair these with our meeting request email templates.
LinkedIn Message Prompts (35–40)
LinkedIn messages need to be shorter and more casual than email. These prompts keep the pitch out of the first touch and lean on shared context instead.
35. Connection request note
Write 3 LinkedIn connection request notes to [name], [title] at [company]. Reference this: [their recent post, shared group, event or role change]. Under 200 characters each. No pitch, no link, no "I'd love to pick your brain".When to use it: The note's only job is getting the accept. Save the pitch for later, or skip it entirely.
36. A comment worth leaving
Here is a LinkedIn post by [name]. Write 3 comments I could leave that add a real point of view, a related example or a thoughtful question. No flattery like "great post". 1 to 3 sentences each.
[paste post]When to use it: A useful comment before a connection request makes your name familiar when the request lands.
37. First message after they accept
[Name] accepted my connection request. Context: [why I connected]. Write a first message under 300 characters that thanks them, asks one genuine question about [their work or post], and does not pitch.When to use it: Start a conversation first. A pitch in message one is the fastest way to get ignored.
38. Follow-up DM after no reply
I messaged [name] a week ago about [topic] with no reply. Write a follow-up DM under 250 characters that shares one useful thing ([resource or insight]) and makes it easy to say "not interested".When to use it: One follow-up is fine. Three in a row on LinkedIn feels like spam.
39. Event or webinar attendee message
[Name] attended [event or webinar] on [topic]. Write a LinkedIn message under 300 characters that references one specific idea from the session, asks what they thought of it, and does not assume they want a demo.When to use it: Shared context is the easiest personalization there is. Use it within a week of the event.
40. Turn a cold email into a LinkedIn DM
Rewrite this cold email as a LinkedIn DM. Make it more casual, under 400 characters, no subject line, no formal sign-off. Keep the signal and the question. Drop anything that sounds like a template.
[paste email]When to use it: Same message, different channel. Pair it with your email sequence so the prospect sees one consistent story.
Send these by hand or through a LinkedIn tool such as HeyReach. SyncGTM does not send LinkedIn messages itself, but it can push enriched leads into HeyReach so your DMs and emails share one list.
Where Live Data Fills the Gaps
ChatGPT is a strong writer and a weak source of prospect facts. It does not have a verified contact database, and it cannot reliably tell you who changed jobs last month or which accounts are hiring this week.
Every prompt above has a slot for a real fact. The better that fact, the better the email. Here is where each kind of fact should come from:
| What the prompt needs | Ask ChatGPT? | Better source |
|---|---|---|
| Verified work email | No, it will guess | Waterfall email finding and verification |
| Hiring signals (prompts 3, 9) | No, may be out of date | Live job openings data |
| Tech stack (prompt 4) | No, may be wrong | Tech stack detection |
| Growth stage (prompts 2, 6) | Rough guess only | Headcount growth and website traffic |
| Turning facts into copy | Yes, this is its strength | ChatGPT, or an AI column in your table |
SyncGTM handles the fact-finding side. In one table, you can add columns like Find Work Email, Company Job Openings, Company Tech Stack, Headcount Growth Rate and Company Website Traffic. Find Work Email runs a waterfall across 50+ data providers.
Then the Write Email with AI column runs a prompt like number 7 on every row, using that row’s real data as the context. You get a personalized draft per lead without pasting anything into a chat window.
Duplicate rows are removed before sync. Finished leads go to Instantly, Smartlead or HeyReach for sending, and to HubSpot, Salesforce, Pipedrive, Attio, Close or Zoho as your CRM record. See SyncGTM pricing for credit costs.
Prefer to work from the terminal? Our guide to Claude Code prompts for sales shows how to run the same research and drafting at list scale.
Mistakes That Make ChatGPT Sales Copy Worse
- No account fact. “Personalize this email” with no input produces flattery about their “impressive journey”. Give it a real signal or skip personalization.
- Letting it invent proof. Tell it not to add numbers, names or results you did not supply, then check every figure.
- No word limit. Drafts run long by default. Set a number, then cut further by hand.
- Sending the first draft. Ask for three versions, pick the best, and rewrite the opener in your own voice.
- Pasting sensitive data. Follow your company’s AI policy, and use a business plan with proper data controls for customer information.
- One prompt for every persona. A VP of Sales and a RevOps manager care about different numbers. Use prompt 15 to split them.
Conclusion
The best ChatGPT prompts for sales are not secret phrases. They are clear instructions with real context: your offer, the buyer, one verified fact, a word limit and a single ask.
Fill in the context block once, save the 40 prompts above, and source prospect facts from live data instead of the model’s memory. When you are ready to run it across a whole list, try SyncGTM free and let the Write Email with AI column do the drafting.