Prompts / Customer Service & Support

ChatGPT prompts for customer success

This prompt pack creates ready-to-use prompts for core customer-success work. Use it when you need structured help without letting AI make promises or invent account facts.

PromptOpen ChatGPTOpen Claude
Create a copy-paste prompt pack for my customer-success work. Build each prompt from the account context below and keep customer facts, commercial terms, and commitments strictly grounded in what I provide.

My customer segment and product: [SEGMENT AND PRODUCT]
My role and responsibilities: [ROLE AND RESPONSIBILITIES]
Account information available to me: [ACCOUNT DATA AVAILABLE]
Success outcomes and health signals we use: [SUCCESS METRICS]
Communication rules, escalation limits, and approved language: [COMMUNICATION RULES]

Produce six distinct prompts, each 120–180 words, for: onboarding plan, business review preparation, low-adoption diagnosis, renewal-risk plan, post-meeting follow-up, and executive sponsor update. Each prompt must tell the AI what account material to request or analyze, specify an output format, distinguish facts from hypotheses, and prohibit invented usage data, pricing, timelines, product commitments, or customer quotes.

For onboarding, require milestones, owners, dependencies, and evidence of value. For a business review, require outcome metrics, trend context, open decisions, and risks. For low adoption, require segmentation by user role and a testable recovery plan. For renewal risk, require risk signals, decision process, mutual action plan, and escalation triggers. For follow-up, require commitments with an owner and due date only if provided. For an executive update, require outcome, risk, ask, and next decision.

Put the prompts under clear task labels. Then add a one-line “Inputs to paste” list below each one. Before answering, check that no prompt directs the AI to make a commercial commitment or present an assumption as account fact. Ask up to three clarifying questions only if a required input is missing.

Fill in

PlaceholderWhat to enterExample
[SEGMENT AND PRODUCT]Describe your customer segment and the product they use.Mid-market HR teams using a workforce scheduling platform.
[ROLE AND RESPONSIBILITIES]State your customer-success role and the tasks you own.CSM for 35 accounts; I own onboarding, adoption reviews, and renewal preparation.
[ACCOUNT DATA AVAILABLE]List the data and notes you can provide, such as usage exports, meeting notes, and support history.Weekly active admins, shift-publishing rate, support tickets, meeting notes, contract dates, and stakeholder map.
[SUCCESS METRICS]List the outcomes and health signals your team uses.Time to first published schedule, manager adoption, reduction in shift conflicts, renewal likelihood.
[COMMUNICATION RULES]Describe approval requirements, tone, escalation paths, and statements you cannot make.Do not promise roadmap dates or discounts; send product defects to Support and commercial terms to the account executive.

How to use

  1. Replace the context fields with your segment, approved data, and guardrails.
  2. Copy one generated task prompt into a new chat with the relevant account materials.
  3. Check that hypotheses are labeled and commitments match approved policy.
  4. Send: “Turn this into a mutual action plan with only confirmed owners and dates: [NOTES].”

Variations

QBR agenda

Use before a quarterly business review with a customer.

Variation
Create a 45-minute QBR agenda for [ACCOUNT] using [ACCOUNT DATA] and these agreed goals: [GOALS]. Attendees and roles are [ATTENDEES]. Return time-boxed sections covering outcomes, usage trends, obstacles, priorities, decisions, and next actions. Identify which claims are facts versus questions to validate live. Include no more than three decisions requested from the customer. Do not invent ROI, benchmarks, or stakeholder opinions. Check that every agenda item supports a stated goal or an identified risk.

At-risk account

Use when signals suggest an account may not renew.

Variation
Assess renewal risk for [ACCOUNT] from [HEALTH DATA], [MEETING NOTES], and [CONTRACT DETAILS]. Return a fact-based risk summary; a table of signal, evidence, confidence, owner, and next validation step; and a 30-day recovery plan. Separate product, adoption, relationship, and commercial risks. Include escalation triggers tied to observed events, not vague concern. Do not assume churn intent, discount authority, or a product commitment. Flag missing decision-maker, renewal date, and success-criteria information.

Onboarding plan

Use when a newly signed customer needs a shared implementation plan.

Variation
Build a customer onboarding plan for [ACCOUNT] adopting [PRODUCT]. Their intended outcomes are [OUTCOMES], stakeholders are [STAKEHOLDERS], and constraints are [CONSTRAINTS]. Produce a milestone table with customer owner, our owner, dependency, completion evidence, and target date only where supplied. Include a kickoff agenda, first-value milestone, adoption checkpoint, and handoff criteria. Do not invent integrations, training capacity, or implementation timelines. Check that each milestone has observable completion evidence and a named owner.

Tips

  • Use the customer’s stated business outcome as the organizing frame; feature usage alone rarely explains account health.
  • Separate observed signals from interpretations in every account plan, especially when a champion has gone quiet.
  • A useful mutual action plan assigns both sides an owner and evidence of completion, not just a list of meetings.
  • Keep product feedback, support issues, and commercial negotiations in separate sections so their owners and promises do not blur.

FAQ

Can AI write customer emails from call notes?

Yes, if it is told to use only confirmed commitments and to label open questions. Do not let it create dates, pricing, or roadmap assurances.

What data should I include for an account review?

Include the customer’s goals, relevant trend data, stakeholder changes, support themes, contract timing, and decisions from recent meetings.

Can AI score account health?

It can organize evidence against your health model, but a score should not replace judgment about data quality and account context.

Related prompts

All Customer Service & Support prompts →

Get a few prompts like this every week

Something Big is a free AI newsletter read by 50,000+ professionals. One email a week with prompts and tools that work, plus what changed in AI and what to do about it.