thinkhard.SI

Prompt repository Customer support & success

Spot churn risk in an account and plan a save

You need risk signals from usage, tickets and notes turned into a plan to keep the customer.

Registry no.
SUP-014
Collection
Customer support & success
Laid out for
Any chat model
Score
10 out of 10
Blanks to fill
7
Added

Customer support & successAny chat

The prompt

Score 10 of 10, Excellent

Use it when You need risk signals from usage, tickets and notes turned into a plan to keep the customer.

  • {{account_name}}
  • {{renewal_date}}
  • {{today}}
  • {{usage_data}}
  • {{tickets}}
  • {{account_notes}}
  • {{approved_levers}}
Full prompt and why it works
Prompt
Review the account information below for signs that {{account_name}} may cancel or not renew, and plan what we should do about it. The review is for me and my manager. Their renewal date is {{renewal_date}}, and today is {{today}}.

Usage data:
"""
{{usage_data}}
"""

Support tickets from recent months:
"""
{{tickets}}
"""

Account notes, emails and meeting notes:
"""
{{account_notes}}
"""

What we are allowed to offer to keep them (training, services, terms, contacts): {{approved_levers}}

Produce:
1. Risk level: high, medium or low, with a one-sentence reason.
2. Signals: a table with Signal, Evidence (with its date), and Weight (strong, moderate, weak). Look for falling usage or active users, a champion or sponsor who left, unresolved or repeated tickets, frustrated language, missed meetings, mentions of budget cuts or competitors, and reduced seats.
3. Likely root cause: the one or two underlying problems the signals point to, such as a team that never finished setup, marked as our hypothesis.
4. Save plan: up to six actions, each with an owner, a date before the renewal, and the signal it addresses. Each action should be specific enough to put in a calendar. Put an honest conversation with the customer first.
5. An opening for that conversation: two or three sentences I could say to start talking about how things are going, without sounding alarmed or salesy.
6. Early signs the plan is working, within the next 30 days.

Keep it under 500 words.

Base every signal on the material above, with its date, and keep facts separate from interpretation. Do not invent usage numbers, feelings or contacts. Offer only the approved levers: do not suggest discounts or free months that are not listed, because a save built on an unapproved offer becomes a bigger problem at renewal.

If the data is too thin or too old to judge, say so, give the risk level as "unknown", and list the information to gather first.

Variables

{{account_name}}
The customer's name.
{{renewal_date}}
When their contract renews or can be cancelled.
{{today}}
Today's date, so the timing of signals and actions is right.
{{usage_data}}
Usage over the last few months: logins, active users, key features used, seats.
{{tickets}}
Recent support tickets with dates, subjects and status.
{{account_notes}}
Your CRM notes, emails and meeting notes, with dates.
{{approved_levers}}
What you may offer to keep them, for example extra training or an executive call. Write "none" if nothing is approved.

Why it works

  • Format. A signals table with dated evidence and a weight for each separates a real trend from one bad week.
  • What to avoid. Only approved levers go into the save plan, because a discount nobody approved turns into a harder conversation at renewal.
  • If unsure. Thin or stale data gets a risk level of "unknown" and a list of what to gather, rather than a confident guess.

How to use it

  1. Press Fill in variables and type your details into each blank. What you type stays in this browser.
  2. Copy the filled prompt into the AI tool you already use, such as ChatGPT, Claude or Gemini, or open it in the Grader or the Prompt Tester here.
  3. Before you rely on it, try it on three or four real examples. The Prompt Tester runs them all at once and shows which ones go wrong.