How to write AI prompts that produce useful, on-brand sales output — every time.
Every effective sales prompt has five core components. Think of them as the scaffolding that tells the AI exactly what role to play, what context to use, and what output you need.
"You are a senior enterprise AE who has closed 7-figure SaaS deals."
"The prospect is a VP of Sales at a 500-person fintech company. They've been using a competitor for 3 years."
"Write a 3-sentence follow-up email after a discovery call where they showed interest but asked for pricing."
"Use bullet points. Keep it under 150 words. No subject line needed."
"Do not mention competitors. Avoid jargon. Sound like a human, not a marketing email."
The best prompts are templates, not one-offs. Use placeholders in brackets to mark the parts you swap out for each deal. This makes prompts reusable, shareable, and faster to use.
[COMPANY NAME]The prospect's company[INDUSTRY]Their vertical (fintech, healthcare, logistics...)[PAIN POINT]The specific problem they mentioned[STAKEHOLDER TITLE]The role of your contact (CFO, VP Sales...)[YOUR PRODUCT]Your solution name or category[DEAL STAGE]Where you are in the cycle (post-demo, pre-close...)Pro tip: The more specific your placeholders, the better your output. "[PAIN POINT]" is fine — "[PAIN POINT: inability to forecast pipeline accurately]" is better.
A long prompt with vague instructions will produce worse output than a short prompt with sharp details. Here's the rule: replace adjectives with facts.
Write a compelling cold email.
Write a cold email to a VP of Revenue at a 200-person B2B SaaS company. They recently raised a Series B. Our product cuts sales cycle length by 30%. Make the subject line a question. Keep it under 100 words.
Help me handle the pricing objection.
The prospect said our price is 40% higher than their current vendor. They're evaluating 3 alternatives. Write 3 responses that acknowledge the gap, pivot to ROI, and ask a question that re-anchors on value, not cost.
Don't start a new chat for every variation. Build on what the AI already knows about your deal in the same conversation. Each follow-up narrows the output toward exactly what you need.
Being too vague about the audience
Always specify job title, company size, and industry. 'Write an email to a prospect' is nearly useless.
Forgetting to set the output format
If you need bullet points, say so. If you need an email with a subject line, say so. The AI will guess — and guess wrong.
No constraints on tone or length
Add word limits and tone guidance every time. Otherwise you get 400-word emails no one will read.
Pasting raw prompt output without editing
AI output is a first draft. Read it, personalize it, remove anything that sounds generic, and add one specific detail from your actual conversation.
Starting over instead of iterating
If the first output is close but not perfect, refine in the same chat. Don't start fresh — the AI already has all the context.
Browse the prompt library and see these principles in action across every deal stage.
Browse Prompts →