Home / Guides / Tool Guide

How to use ChatGPT for writing follow-up emails

Short answerWrite the sequence once with ChatGPT, then send it every time. The money is in the follow-ups, not the first email: in a study of 7.5 million sent emails, steps two through six produced 58.6 percent of all replies. Three to five emails is the range. Past that you are mostly generating unsubscribes.

How to do it

1

Build a sequence, not a single email

Belkins analyzed 7,530,489 emails over twelve months. The first email produced 41.4 percent of replies. Steps two through six produced the other 58.6 percent. Step three alone drove 35.6 percent of all meetings booked from email, more than steps one and two combined. If you send one quote follow-up and stop, you are leaving most of the response on the table.

2

Stop at three to five emails

The same analysis puts the practical range at three to five steps, with returns falling off fast after step five. Per-step reply rates drop from about 0.59 percent at step one to roughly 0.28 percent by step six. Build five and stop. A ten-email sequence is not five times better, it is just more work and more people annoyed at you.

3

Give ChatGPT the context it cannot guess

OpenAI's guidance is role, reference material, defined output. So: 'You write follow-up emails for a commercial plumbing contractor. Context: we quoted $8,400 for a backflow replacement on July 30, spoke to the facilities manager, no response since. Write email two of a four-email sequence. Under 90 words. One specific ask. No pleasantries.' Do that and the output is usable.

4

Make each email carry something new

The reason most follow-ups fail is that they say 'just checking in,' which gives the reader nothing to react to. Tell ChatGPT what new thing each email brings: a similar job you finished, a lead time that is about to change, a price that holds until a date, an answer to the objection you expect. One new fact per email.

5

End with a question they can answer in four words

'Let me know your thoughts' gets nothing. 'Do you still want this done before September?' gets a yes or a no. Put the exact call to action in the prompt rather than hoping the model picks a good one. Ask it for three subject line options per email while you are there.

6

Add the legal bits ChatGPT will leave out

Commercial email in the US has to carry a valid physical postal address and a working opt-out mechanism that stays live for at least 30 days, and you have to honor opt-outs within 10 business days. Subject lines cannot mislead about the contents. ChatGPT does not add any of that on its own. Put it in your template, not in the prompt.

7

Save the sequence as a template and stop rewriting it

The point of drafting with ChatGPT is that you do this once. Paste the five emails into whatever you send from, swap the variable pieces per customer, and send. If you are asking ChatGPT to write a fresh follow-up every single time, you have automated nothing.

Where this goes wrong

The parts most guides skip, and the ones that actually cost you money:

Free tools for service businesses

Quotes, invoice follow-ups and review requests. No signup.

Open the free tool

Common questions

How many follow-up emails should I actually send?

Three to five total, including the first email. Belkins' analysis of 7.5 million emails puts the sweet spot at three to five steps, with diminishing returns setting in fast after step five.

Does following up really change anything?

Yes, substantially. In that dataset the first email accounted for 41.4 percent of replies and steps two through six accounted for 58.6 percent. Step three alone drove 35.6 percent of all meetings booked.

What has to be in a commercial email legally?

In the US: accurate header information, a subject line that does not mislead about the contents, a valid physical postal address, and a clearly displayed opt-out mechanism that works for at least 30 days. Opt-out requests must be honored within 10 business days.

Why do ChatGPT's follow-ups sound generic?

Because the prompt is generic. Give it the role, the actual quote amount, the date, what was discussed, and a hard word limit and format. OpenAI's own guidance is role, reference material, defined output.

Should each follow-up repeat the original offer?

Briefly, yes, but the email needs a new reason to exist. One new fact per email: a comparable job, a lead time, a deadline on the price, or an answer to the objection you expect.

Sources

  1. Sales follow-up statistics in B2B — BelkinsStudy of 7,530,489 emails over 12 months: step 1 produced 41.4% of replies and steps 2-6 produced 58.6%; step 3 drove 35.6% of email-sourced meetings; per-step reply rate fell from 0.59% at step 1 to 0.28% at step 6; practical sweet spot is 3 to 5 steps.
  2. 15 U.S. Code 7704: Other protections for users of commercial electronic mail — Cornell Law School LIICommercial email may not use materially false or misleading header information or subject lines, must include a clearly and conspicuously displayed functioning opt-out mechanism valid for at least 30 days, must honor opt-outs within 10 business days, and must include a valid physical postal address.
  3. Prompt engineering guide — OpenAIRecommended tactics include explicit role and workflow guidance, supplying relevant reference context, few-shot examples, and defining output structure and format.

Related guides

Want this built for you instead?

Tell us the job you keep doing by hand. If we can automate it, we will quote it. If we cannot, we will tell you that too. No calls, everything in writing.

Got it. We will reply to that address shortly with what it would take.
Nexus AI Solutions · Circleville, Ohio · Operational AI for service businesses · Free tools · All guides