AI Prompts for Writing Sales Emails

Picture this: you're a sales rep at a 30-person software company. You've got a list of 50 prospects that your SDR marked as "interested" three weeks ago. Now it's your job to follow up. You open a blank email draft and realize you have absolutely no idea how to write something that doesn't sound like every other sales email these people have already deleted.

That's the exact problem ai prompts sales emails workflows are built to solve—not by generating blasts, but by helping you write emails that feel like they were written by a human who actually knows the prospect's situation.

Before: The Weak Prompt

The most common mistake: "Write a sales email for [product] to [prospect type]."

What comes back is the email equivalent of a form letter. Generic opening, vague value prop, weak CTA. If you've ever gotten a cold sales email that started with "I wanted to reach out because..." that's AI (or a lazy human) running this exact prompt.

The output isn't usable because it has no specificity. It doesn't know who the prospect is, what their situation is, what they've already heard from you, or what specifically your product does for their exact role.

Why It Fails

A weak sales email prompt fails for two consistent reasons. First, it doesn't include context. AI needs to know: who is this person, what do you know about them, and what's the specific situation you're opening with? Without that, it defaults to generic value propositions.

Second, it doesn't specify the email's job. A first touch email, a follow-up email, and a breakup email are all "sales emails" but they serve completely different functions and need completely different structures.

⚠️ Common mistake: Asking AI to write "a good sales email" without defining what stage of the sequence this is. A great first-touch email has a completely different structure than a great third-touch email. The first touch needs to earn attention. The third touch needs to create urgency or provide new information. Define the stage before you define anything else.

After: The Improved Prompt

Here's a sales email prompt that produces usable first drafts:

Write a sales email from [your name/title] at [company] to [prospect name/title] at [prospect company].

Context:
- What I know about them: [2-3 specific facts—company news, LinkedIn activity, job posting, technology they use, etc.]
- My product/service: [one sentence on what it does]
- Specific problem it solves for their role: [one sentence]
- Stage of outreach: [first touch / follow-up #X / post-demo / breakup]
- Previous touchpoints: [list if any]

Email requirements:
- Subject line: under 7 words, no clickbait, create curiosity or relevance
- Opening: reference the specific context above—NOT a compliment about their company
- Body: 3-4 sentences maximum on the specific problem and how we solve it
- CTA: one specific action, not "let me know if you're interested"
- Tone: direct, peer-to-peer—not salesy

No bullet points. No bold text. Write like a human.

What this does: The "no compliment about their company" instruction is the most important constraint in the entire prompt. AI sales emails default to openers like "I've been following [Company] and love what you're doing." Every prospect has seen that 40 times and knows immediately it's a template.

⚡ Pro tip: For the opening line, run this separately: "Write 5 different opening lines for a sales email to [prospect type] that reference [specific context]. Each opening must be different in structure—not just different words. Avoid openers that start with 'I' or 'We.'" Pick the one that would most make you stop scrolling if you received it.

Breaking Down Each Element

The subject line instruction ("under 7 words, no clickbait") sounds restrictive, but shorter subject lines consistently outperform longer ones in sales contexts. The goal isn't intrigue—it's relevance. "Question about [their company's recent action]" or "[mutual connection] suggested I reach out" outperform anything that sounds like a marketing subject line.

The 3-4 sentence body is harder than it sounds. Most first drafts are too long. After the AI generates the email, run this: "Cut this email to 4 sentences without losing the core message. The most important sentences to keep are the opening context hook and the CTA."

The specific CTA is where most AI sales emails fall apart. "Let me know if you're interested" puts all the work on the prospect. A specific CTA—"Would Tuesday or Wednesday afternoon work for a 20-minute call?"—reduces friction and increases reply rates.

⚡ Pro tip: For follow-up emails, use this variant: "Write email #[X] in a sales sequence to [prospect] who hasn't responded to my previous [X-1] emails. My previous emails covered: [summarize]. This email should either provide new information, take a different angle, or directly acknowledge the silence—don't just repeat what I've already said."

Variations for Different Sales Contexts

Post-demo follow-up:

Write a follow-up email after a sales demo with [prospect name/title] at [company]. The demo covered: [key points shown]. They seemed most interested in: [specific feature or use case]. Their concerns were: [any objections or questions]. Next step we agreed on: [what was promised].

This email should: confirm the next step, address their main concern in one sentence, and include one piece of relevant content (case study, ROI data) they can share internally. Under 150 words.

Re-engagement email (prospect went cold):

Write a re-engagement email to [prospect] who went dark after [last touchpoint, X weeks ago]. I don't know why they stopped responding. The email should: acknowledge the gap without being passive-aggressive, offer a new reason to reconnect (new feature, customer story, relevant industry event), and make it easy to either re-engage or close the loop. Tone: casual, zero pressure.

What this does: The re-engagement email is the one most reps write worst, because they're either too apologetic or too aggressive. This prompt structure produces the middle path: direct acknowledgment of the silence with a genuinely new hook.

⚡ Pro tip: Build a "persona + context" snippet for each ICP you sell to. Something like: "[VP of Operations at 100-500 person manufacturing companies] typically cares about: [list]. Common objections from this persona: [list]. Language they respond to: [list]." Paste this snippet into every AI sales email prompt for this segment. Your emails become consistently more relevant without requiring you to research each prospect individually.

Save and Reuse This

Using AI to Adapt Emails for Different Seniority Levels

One underused application of sales email prompts: adapting the same core message for different seniority levels within the same company. A first-touch email to a VP Operations needs a different framing than the same product pitched to a Director of Supply Chain—even at the same company, selling the same thing.

I have this sales email written for [VP/C-level]: [paste email]. Rewrite it for [Manager/Director level] at the same company. Key changes needed: (1) shift from business outcome framing to operational/day-to-day framing, (2) make the specific pain point more tactical and immediate, (3) adjust the CTA to something appropriate for someone who influences but doesn't make the final decision. Keep the same length and tone.

What this does: Produces persona-appropriate variants from a single core message without having to rethink the entire email from scratch. A manager doesn't care about the same things a VP cares about—and an email written for the wrong level of the org gets deleted fast.

A sales email prompt library organized by persona, stage, and context is one of the highest-value assets a sales or marketing team can build. When a new SDR joins, they're not starting from scratch—they're pulling from tested prompts with notes on what works.

PromptABCD is the right home for this library. Tag prompts by email stage (first touch, follow-up #2, post-demo, breakup), by ICP, and by deal size. The sales team that invests in this prompt library in month one gets compounding returns as those prompts get refined over time.

Testing and Improving Your Sales Emails Over Time

The gap between a mediocre sales email prompt and a great one is usually 3-4 iterations of refinement based on real response data. But most teams don't close that loop systematically.

Here's a prompt that helps:

Here are 3 sales emails I sent to [ICP] last month, with their actual reply rates: [paste emails with reply rates]. Analyze what's working and what isn't. For the lowest-performing email, give me 3 specific rewrites—not the same email with different words, but genuinely different approaches (different hook, different structure, different CTA). Explain what hypothesis each rewrite tests.

What this does: Framing each rewrite as a "hypothesis" forces a testing mindset instead of a "try something different and hope" mindset. You're building knowledge about your specific ICP with each iteration.

⚡ Pro tip: Build a "swipe file" prompt after any sales email gets a strong reply rate: "This sales email got a [X]% reply rate with [ICP]: [paste email]. Analyze what elements are driving this performance. Write a brief rubric I can use when reviewing future AI-generated sales emails for this ICP, based on what this email is doing well." That rubric becomes part of your prompt for every future email to that segment.