Write Sales Emails with ChatGPT

Email remains a core channel for B2B sales. Yet writing a cold outreach, a follow-up, or a sales proposal takes time and requires a subtle balance between personalization and efficiency. ChatGPT speeds up this process without compromising quality, provided you keep one thing in mind: the model knows nothing about your prospect until you tell it.

The four scenarios to cover

Cold outreach is the hardest to write because it must convince someone who isn’t expecting you. ChatGPT helps you find the right hook and personalize the message for the prospect, but you must set the length constraint: without it, the model will produce a message that’s far too long for anyone to read.

“Write a cold outreach email for [prospect name], [title] at [company]. Their company does [activity] and has [size]. Our offer: [one-sentence description]. The email must be 5 lines maximum, direct, and propose a 15-minute call. No empty flattery.”

Follow-ups after an initial contact follow a different logic: you’ve already spoken, and the goal is to move the deal forward one step without seeming pushy. Addressing the objection raised during the conversation is what separates a useful follow-up from a simple reminder.

“Write a follow-up email after our phone call on Tuesday with [first name]. We discussed [topic]. They seemed interested in [specific point] but hesitated about [objection]. Propose a concrete next step.”

The email that accompanies a sales proposal is read first, often on mobile, and determines whether the attachment gets opened. It must state the amount, key benefits, and validity period without paraphrasing the document.

“Write the cover email for a sales proposal to [client]. Amount: €[X]. Key benefits of our offer: [3 points]. Validity period: 30 days. Tone: professional and confident.”

Finally, post-sale follow-up is the most profitable yet most often overlooked. Thirty days after delivery, a short message checks satisfaction, opens the door to a follow-up meeting, and reminds them of your referral program.

“Write a follow-up email 30 days after delivering [product/service] to [client]. Objectives: check satisfaction, propose a follow-up meeting, mention our referral program.”

Build a template library

Writing every email from scratch means endlessly redoing the same work. Instead, create reusable templates in a dedicated project where your tone and length instructions apply automatically. Name each template clearly—“Outreach – Manufacturing Sector,” “Follow-up D+7,” “SMB Proposal”—and keep the variables highly visible: [first name], [company], [offer], [amount], so no send goes out with a missing field.

There’s one step people almost always skip: after about ten sends, analyze open and reply rates, then ask ChatGPT to adjust the underperforming templates. A template is never universally good; it’s good for a specific sector and time period. Without measurement, you’ll keep using a subject line that no longer gets opens for months. Here’s the prompt that produces a ready-to-use template on the first try.

“Create a cold outreach email template for SMBs in the [X] sector. Include these variables: [first name], [company], [specific pain point]. The email must be 100 words maximum, with a catchy 6-word subject line. Provide 3 subject line variants.”

Personalize at scale

Personalization is what separates an email that gets read from one that gets deleted, yet it’s the first casualty when time is short. ChatGPT lets you personalize every message without spending twenty minutes on each, as long as you share what you know about the prospect before asking it to write. A few minutes of research—a published article, a funding round, a job posting—is enough.

“Prospect context: [name] published a LinkedIn article last week about [topic]. Their company just raised €[X]M. They’re hiring a [role], which suggests they’re expanding [activity]. Use these details to personalize the cold outreach email.”

Notice the reasoning in that last sentence: hiring for a specific role is interpreted as a signal of growth. This kind of deduction, supplied by you, enables the model to write a relevant hook rather than generic flattery.

Best practices

A few habits preserve these results. In B2B, a cold outreach email should be just a few lines: explicitly ask for conciseness, or the model will fill the space. Craft the subject line carefully—it alone determines whether the email gets opened—and always ask for multiple variants. Resist the temptation of volume: a personalized email sent to twenty people often outperforms a generic one sent to two hundred. Finally, check CNIL rules (France’s data protection authority) before any campaign. For professionals, email outreach can proceed without prior consent if it relates to their profession, provided you inform them and allow them to opt out easily and free of charge with every send. For consumers, prior consent is required, except for existing customers when you’re offering similar products or services. No productivity gain offsets a formal notice of violation.

Your turn

Set up the routine. Identify the three types of emails you send most often, then build a template for each with ChatGPT. Test them on five real recipients—not colleagues—and measure the reply rate before iterating.

Official sources: Prompting (ChatGPT documentation) · Projects in ChatGPT (OpenAI Help Center) · Commercial outreach by email, SMS/MMS, and automated calling (CNIL)

Key takeaways

• ChatGPT excels at writing sales emails when given precise context.
• Build a template library in a project to boost efficiency.
• Personalize every email with prospect-specific information.
• Measure results (open rates, reply rates) and iterate on templates.
• Follow CNIL rules: opt-out available with every send in B2B; prior consent required in B2C.