Skip to content
If it doesn't work, you don't pay. 30 days.
Implementa.

Using ChatGPT in the enterprise · Guide 4 of 5

How to use ChatGPT to prospect (cold email that doesn't look like AI)

Using ChatGPT to prospect works when the prompt is paired with real prospect data and ends up going through a human. Without the first, the emails look written by AI. Without the second, they were written by AI. Either kills the reply rate.

See it in production → AI Sales SystemBack to the pillar

The 4 inputs the prompt needs (without them it doesn't work)

Without all four inputs together, the prompt produces dressed-up templates. Missing even one is the difference between a 0% and a 4% reply rate:

  1. Product context. What you sell, who you sell to, concrete value proposition (not "we help companies grow").
  2. Real prospect profile. Title, company, industry, size, recent signals (funding, hiring, press mention).
  3. Specific campaign angle. Not "introduce the company", but "why this specific prospect right now".
  4. Tone and format. Language, register, max length, exact CTA.

Tested prompt templates

Structure that works in US B2B mid-market:

  • System prompt: role, brand, value prop, tone, length (60-100 words), format (specific CTA).
  • User prompt: prospect data (name, title, company, industry, recent signal) + concrete angle + example of the desired CTA.
  • Post-validation: detect and strip clichés, ensure mention of at least 2 specific data points.

How to connect ChatGPT to the prospect's public data

  • Manual (low volume). Copy LinkedIn + press section of the site + latest relevant post → into the prompt.
  • Semi-automated (Make/Zapier + Clay). Clay enriches + Make passes to ChatGPT + returns a ready-to-send email.
  • Fully automated (custom). Pipeline with legitimate scraping + LLM + send via Instantly/Smartlead.

When to jump to a real AI SDR

Using ChatGPT for prospecting is phase 0 — useful for low volume and for learning. When you outgrow it:

  • Volume >150 emails/week.
  • Need to manage deliverability and warm-up.
  • Reporting of pipeline generated vs. emails sent.
  • Bidirectional CRM integration.
  • Systematic A/B testing of copy and segments.

Measure replies, not opens

Open rates have become unreliable thanks to privacy protections. The honest metrics for outbound:

MetricReasonable B2B read
Positive reply rate1-3% cold, 5-12% well segmented
Meetings bookedDepends on volume; ratio of meetings/sends
Qualified pipeline generatedThe final metric
Bounce rate<2% healthy
Spam complaints<0.1%

Frequently asked questions

From a new domain: start with 10-20/day for 2-3 weeks (warmup), ramp progressively to 50-100/day per inbox. From a warmed domain: 50-150/day per inbox as a reasonable max. Going beyond is playing with reputation and deliverability drops fast. Any vendor saying "send 500/day from a new domain" doesn't know what they're selling.

B2B: Apollo, Clay, ZoomInfo or LinkedIn Sales Nav are the standard sources. For enrichment (role, company, signals): Clearbit or Crunchbase. Rule: prompt quality is directly proportional to input data quality. Generic data = generic emails = 0% reply rate, no matter how you tune the prompt.

Email for volume and scale; LinkedIn for senior and ABM. The combination works better than each alone: LinkedIn to "be on their radar", email to "open the conversation". LinkedIn-only leaves volume on the table; email-only leaves relevance.

Free AI Impact Plan

The guide is generic. Your plan isn't.

Tell us about your company and we'll ship back a diagnosis with priorities, numbers and what to implement first. No sales call, no charge.

Keep going with these

How to use ChatGPT in a real company (not in a LinkedIn post)How to use ChatGPT for sales: prompts, flows and where it breaksHow to use ChatGPT in companies: policy, cases and typical mistakes
How to use ChatGPT to prospect (cold email that doesn't look like AI) · Implementa