Skip to content
Kit IA

← All prompts

Clients & Operations · Updated July 3, 2026

ChatGPT prompt for re-engaging an at-risk client

This is the prompt we run when a client shows exit signals: a silence that settles in, cancelled check-ins, a renewal getting close. Its core rule: separate what the signals show from what you assume, before writing a single message.

The prompt to copy

Client {name} shows churn risk. Signals: {list what you've observed}. History: {one paragraph}.
Produce:
1. An honest read of what's likely going on — clearly separate what the signals show from speculation.
2. What we'd need to learn before reacting, and how to learn it.
3. A short outreach message that opens the real conversation — curious, not panicked, not salesy. Under 100 words.
4. If they do want out: what a graceful exit looks like that preserves the relationship and referrals.

Replace the fields in curly braces (the client name, the signals you observed, one paragraph of history), then list the signals exactly as you noted them: dates, facts, no interpretation.

When to use it

As soon as a client shows exit signals: reduced usage, missed calls, a complaint, a contract ending soon. The right moment is early, while the signals are still weak, not the week the notice period starts. Six weeks before a renewal, there is still time for a real conversation.

What you get back

  1. An honest read of the situation, facts on one side, assumptions on the other.
  2. The list of what to verify before reacting, and how.
  3. A short opening message, curious, not panicked, not salesy.
  4. The graceful-exit scenario, if the client really does want out.

Real example: raw signals → opening message

Input: the signals noted by the account manager of an 8-person agency, about a B2B SaaS client on a content retainer. An excerpt, as-is:

- radio silence for 3 weeks, one-line replies before that
- May and June check-ins cancelled by their marketing lead, never rebooked
- retainer renewal in 6 weeks
- last article batch approved without a single comment (unusual: they annotate everything)

The honest read, one line that sets the tone: “What the signals show: the disengagement is real, recent, and concentrated on one contact. What they do not show: the cause. A budget review, a swamped contact, or another agency are all, at this point, speculation.”

And the generated opening message:

Hi Jordan,

I noticed our last two check-ins fell through, and I would rather ask directly than guess: is the retainer, in its current shape, still doing what you need?

If your priorities shifted this quarter, I would genuinely like to understand what changed, so we can adjust on our end. Would you have 20 minutes this week or next?

And if the timing is bad, just say so.

Sam

Notice what the message does not do: no panicked discount, no “we’d love to walk you through our new offerings”. One direct question the client can take seriously. The human review, checking the name and the tone, takes two minutes.

Prompt C8 ships inside the kit, as part of the Clients & Operations prompt library, ready to copy with its fill-in fields.

Frequently asked questions

Does this prompt work with Claude or Gemini?
Yes. It relies on no ChatGPT-specific feature: paste it as-is into Claude or Gemini and the behavior is the same.
Why does the prompt separate signals from speculation?
Because reacting to a wrong guess makes things worse. A swamped client who receives a panicked retention email starts wondering what you are worried about. The prompt forces the split between what the facts show and what you assume, then lists what to verify before you write.
Is the outreach message ready to send?
Almost. It is a solid first draft, but read it over: check the name, the context, the tone, and rewrite anything that does not sound like you. Two minutes of review before a message this sensitive is not optional.

Related prompts