Copy & deploy · 11 prompts

11 Claude prompts for customer success.

Eleven production-grade Claude prompts for Customer Success professionals at B2B companies. Each one is built around a real CS job, from onboarding plans and QBR decks to health-score triage, churn saves, renewal prep, and expansion plays. Copy a prompt, swap in your account details, and run it.

ONBOARDING

Onboarding and time-to-value

01

30-60-90 onboarding plan from the sales handoff

Use when: a new account moves from Closed Won to your queue and you need a launch plan before the kickoff call
You are my Customer Success co-pilot for a B2B [PRODUCT CATEGORY] platform. Build a 30-60-90 day onboarding plan for a new account.

Here is the sales handoff context:
- Account: [COMPANY NAME], [INDUSTRY], [HEADCOUNT / ARR]
- Why they bought (from sales notes): [PASTE NOTES]
- Their stated success metric: [E.G. CUT TICKET RESOLUTION TIME 20%]
- Champion: [NAME, ROLE]. Other stakeholders: [LIST]
- Technical setup required: [INTEGRATIONS / SSO / DATA IMPORT]
- Known risks or red flags from the deal: [LIST]

Produce:
1. The single first-value milestone they should hit, and by what date, that proves the purchase decision was right.
2. A week-by-week plan for days 1-30, then 31-60, then 61-90, with owner for each task (me, the customer, or shared).
3. The exact data, access, or decisions I need to request from the champion in the kickoff, written as a short numbered ask.
4. Two leading indicators I should watch in each phase to know if onboarding is on track or stalling.

Flag anything in the handoff that is vague or missing and tell me what to confirm with sales before kickoff.
02

Kickoff agenda and pre-read

Use when: you have a kickoff scheduled and want a tight agenda plus a one-page pre-read the customer will actually read
Write a 45-minute kickoff agenda and a one-page customer-facing pre-read for this new account.

Context:
- Product: [PRODUCT], used by [USER PERSONAS]
- Customer goal: [GOAL]
- Attendees on their side: [NAMES, ROLES]
- Attendees on my side: [NAMES, ROLES]
- What must be true by end of call: [E.G. ADMIN ACCESS GRANTED, SUCCESS METRIC AGREED, NEXT 2 MEETINGS BOOKED]

Agenda rules: time-boxed, outcome per segment, no live demo unless it drives a decision. Reserve the last 10 minutes for mutual commitments and dates.

Pre-read rules: plain language, no internal jargon, under 350 words, structured as Why we are here, What good looks like in 90 days, What we need from you, Who does what. Write in a warm, confident voice. Do not use em dashes.
HEALTH AND RISK

Health scoring and risk triage

03

Account health read from mixed signals

Use when: you have usage data, support history, and meeting notes for one account and need a fast, honest health verdict
Act as a skeptical Customer Success analyst. Read the following signals for one account and give me a clear-eyed health assessment.

Usage data: [PASTE: logins, active seats vs licensed, feature adoption, trend over last 90 days]
Support history: [PASTE: ticket count, severity, sentiment, open issues]
Relationship notes: [PASTE: last 2-3 meeting summaries, champion status, exec sponsor presence]
Commercial: [RENEWAL DATE, ARR, contract terms]

Do this:
1. Give a health rating of Green, Yellow, or Red, and state the single most important reason for it.
2. Separate real risk signals from noise. Call out where the data looks fine on the surface but worries you, and why.
3. List the top 3 risks ranked by likelihood times impact on renewal.
4. Recommend the next 2 concrete actions with owners and timing.

If the signals contradict each other, say so plainly and tell me which signal to trust most for this kind of account.
04

Churn-risk early-warning scan across the book

Use when: you manage a portfolio and want to spot at-risk accounts before they show up in a renewal forecast
I am going to paste a table of accounts from my book of business. Help me triage churn risk across the portfolio.

Columns: [ACCOUNT, ARR, RENEWAL DATE, ACTIVE SEATS / LICENSED, 90-DAY USAGE TREND, OPEN TICKETS, CHAMPION STATUS, LAST QBR DATE]

Data:
[PASTE TABLE]

Produce:
1. A ranked at-risk list, highest risk first, with a one-line reason per account.
2. The 3 patterns you see repeating across at-risk accounts, since those point to a fixable systemic issue, not just one-off saves.
3. A suggested triage tier for each: Save now, Watch, or Healthy, with the trigger that would move a Watch account to Save now.
4. Which 5 accounts I should personally touch this week and the opening line for each outreach.

Be direct. Do not soften a Red just because the ARR is large.
RENEWALS AND EXPANSION

Renewals, QBRs, and expansion

05

QBR deck outline grounded in outcomes

Use when: a quarterly business review is coming up and you want a value-led narrative, not a feature recap
Build the outline for a quarterly business review with [CUSTOMER]. The goal is to prove value and set up the renewal conversation, not to demo features.

Inputs:
- Their original success metric: [METRIC] and where it stands now: [CURRENT NUMBER]
- Usage and adoption highlights this quarter: [PASTE]
- Wins and customer quotes if any: [PASTE]
- Open issues or unmet expectations: [PASTE]
- Their business priorities for next quarter: [PASTE]

Produce a slide-by-slide outline with the headline for each slide written as a takeaway, not a topic. Include:
1. A results slide that ties usage to their business outcome in their language.
2. An honest slide on what did not go well and how we are fixing it.
3. A forward plan slide with 2-3 goals for next quarter.
4. A natural bridge to renewal and one expansion idea, framed around their priorities.

Give me the talk track for the results slide and the renewal bridge. Keep it executive-ready. Do not use em dashes.
06

Renewal risk brief and negotiation prep

Use when: you are 60-90 days from a renewal and need to walk in knowing the risks and your asks
Prepare me for a renewal conversation with [CUSTOMER], renewing on [DATE] at [CURRENT ARR].

Context:
- Value delivered this term: [PASTE OUTCOMES, METRICS]
- Relationship state: [CHAMPION, SPONSOR, ANY TURNOVER]
- Adoption: [SEATS, FEATURES, TREND]
- Known headwinds: [BUDGET PRESSURE, COMPETITOR, REORG, ETC.]
- My goal for this renewal: [FLAT, UPLIFT %, MULTI-YEAR, EXPANSION]

Produce:
1. A renewal risk brief: likelihood of renewal, the 3 biggest threats, and the proof points that counter each.
2. The value case I should lead with, stated as their ROI in their terms.
3. Three likely objections and a crisp response to each, including the one I am least prepared for.
4. My opening position, target, and walk-away, plus a concession ladder if they push on price.
5. The one question that will surface their real intent early in the call.
07

Expansion play from usage patterns

Use when: an account is healthy and you want to identify a credible, customer-first expansion path
Find the strongest expansion opportunity in this healthy account and pressure-test it before I pitch it.

Account context:
- Current plan and usage: [PASTE: tier, seats, features used vs available]
- Teams or use cases not yet on the platform: [LIST]
- Stated goals and roadmap: [PASTE]
- Budget cycle and decision process: [WHAT I KNOW]

Produce:
1. The single best expansion play, whether seats, tier, module, or new team, and the usage evidence that makes it credible.
2. The customer-first framing: how this solves their problem, not how it grows my number.
3. The internal champion most likely to back it and why.
4. A 3-step sequence to introduce it without making the relationship feel transactional.
5. The signal that tells me to hold off because the timing is wrong.

If the data does not support a real expansion, tell me that instead of inventing one.
CUSTOMER COMMUNICATION

Customer-facing communication

08

De-escalation reply to an upset customer

Use when: a customer sends a frustrated or escalating email and you need a reply that holds the relationship
Draft a reply to this upset customer. The relationship matters and I do not want to be defensive or dismissive.

Their message:
[PASTE EMAIL]

What is actually true on our side:
[PASTE: what happened, what we can and cannot do, timeline for a fix]

Constraints:
- Acknowledge the impact before explaining anything.
- Do not over-apologize or admit fault we do not own, but do not deflect either.
- Be specific about the next step, owner, and date.
- Match a calm, senior, human tone. No corporate filler.
- Keep it under 180 words.

Give me two versions: one warmer, one more direct. Then list the one thing I should confirm internally before I hit send. Do not use em dashes.
09

Feature-request response that keeps trust

Use when: a customer asks for something that is not on the roadmap or will not ship soon
Help me respond to a customer feature request without damaging trust or over-promising.

The request: [PASTE]
Why they want it (the underlying job): [WHAT I KNOW OR SUSPECT]
Reality: [NOT PLANNED / ON ROADMAP FOR Q? / NEEDS MORE DEMAND / WORKAROUND EXISTS]
Available workaround if any: [DESCRIBE]

Produce:
1. A reply that restates their underlying need so they feel heard, gives the honest status, and offers the best available path today.
2. If there is a workaround, clear steps to use it.
3. A line that keeps the door open and tells them how their input affects prioritization, without promising a ship date.
4. A short internal note I can log to our product team capturing the request, the account, and the business impact.

Keep the customer reply honest and warm. Do not use em dashes.
INTERNAL AND VOICE OF CUSTOMER

Internal reporting and voice of customer

10

Internal account save plan

Use when: an account goes Red and you need a focused recovery plan to align your team and manager
Build an internal save plan for a Red account so my team and manager are aligned on the recovery.

Situation:
- Account: [NAME, ARR, RENEWAL DATE]
- Why it went Red: [ROOT CAUSE, NOT JUST SYMPTOMS]
- What we have tried: [PASTE]
- Stakeholders and their current sentiment: [LIST]
- What we can credibly offer: [E.G. EXEC SPONSOR, SERVICES HOURS, ROADMAP COMMITMENT, COMMERCIAL FLEX]

Produce a save plan with:
1. The root cause stated in one sentence, separating what we caused from what we cannot control.
2. A 2-week action plan with daily-level owners and a single defined success criterion.
3. The internal escalation: who I need from product, exec, or leadership, and the exact ask for each.
4. A realistic probability of save and the trigger point at which we shift from saving to managing a graceful exit.

Be honest about whether this account is savable. Do not give me false hope.
11

Voice-of-customer synthesis for product and leadership

Use when: you have a pile of customer feedback and need to turn it into themes leadership will act on
Synthesize this raw customer feedback into a voice-of-customer brief for product and leadership.

Sources (mixed): [PASTE: QBR notes, support tickets, churn reasons, survey verbatims, call snippets]
Segment or filter if relevant: [E.G. ENTERPRISE ONLY, LAST 90 DAYS]

Produce:
1. The top 5 themes ranked by frequency and by revenue or retention impact, with a representative customer quote for each.
2. For each theme, separate what is a product gap, a support gap, an onboarding gap, or an expectation-setting gap, since the owner differs.
3. The 2 themes most tied to churn or stalled expansion, with the evidence.
4. One slide-ready summary leadership can absorb in 30 seconds.
5. The single change that, if shipped, would move retention the most, and the confidence level behind that claim.

Do not flatten dissenting feedback into a false consensus. If a theme is loud but low-impact, say so.
FAQ

Frequently asked questions

How should we handle PII when feeding customer data into these prompts?
Treat the prompt window like a shared doc. Before pasting a ticket thread, health score export, or CSM call transcript, strip the fields you do not need for the task: full credit card numbers, SSNs, raw API keys, dates of birth, home addresses. Account names, contract values, usage metrics, and stage notes are usually fine if your contract with Anthropic is in place and your workspace has zero data retention turned on for the API. If you are on a free or consumer plan, assume the input can be used for model improvement and redact accordingly. For regulated data (HIPAA, PCI, anything covered by a customer DPA), route the work through the Anthropic API with a signed BAA or your enterprise admin's approved workspace, not a personal Claude.ai login.
Which Claude model should a CS team default to?
For day to day CSM work, Sonnet is the right default. It is fast enough for live ticket triage, QBR prep, and renewal brief drafting, and the quality is more than enough for those tasks. Move to Opus when you are doing the heavier lifting: synthesizing six months of customer calls into a churn root cause analysis, building a playbook from scratch, or writing the strategy deck for a top ten account. Haiku is the one to reach for when you are running high volume automations (auto tagging tickets, summarizing every closed conversation overnight) where cost per call matters more than nuance.
Can Claude work directly with Gainsight, Catalyst, ChurnZero, or Vitally?
Yes, in two ways. The lightweight path: export the view you care about (an at risk segment, a CTA list, a timeline) as CSV or copy the JSON from the API, paste it into Claude, and ask for the synthesis you want. The heavier path: stand up an MCP server or a simple middleware that lets Claude pull customer records, health scores, and playbook CTAs on demand. Most CS Ops teams start with paste in workflows for a quarter, see which prompts get used daily, then wire those specific flows up to the CSP via API so the analyst is not the bottleneck. Same pattern works for HubSpot Service Hub, Intercom, Zendesk, Front, and Planhat.
How do we keep the voice consistent across CSMs when everyone is using Claude?
Two things make the biggest difference. First, write a one page voice brief: how your company talks to customers (warm but direct, no jargon, never apologetic for product gaps, etc.), with three real before and after examples. Paste that brief into every prompt as the system message or top of the user message. Second, build a shared prompt library inside your CS wiki (Notion, Guru, Confluence) so a new CSM is not inventing their own version of the renewal email prompt. The prompts in this page are a starting kit; the voice brief is what makes the output sound like your company instead of generic SaaS.
When should a CSM not let Claude write the renewal email?
Three cases. One: the renewal is contested, meaning the customer has already flagged churn risk, raised pricing concerns, or escalated to legal. That conversation needs a human who can read the room and make trade offs in real time. Two: the contract value or strategic importance is high enough that a slightly off note costs you the account. Top ten customers, design partners, logos you put on the website. Three: the relationship is built on a specific personal history (the CSM was at the customer's wedding, they swap birthday texts) where AI generated phrasing will read as cold or lazy. In all three cases, use Claude to draft the strategy memo or talking points, then write the actual email yourself.
What is a reasonable way to measure whether these prompts are actually helping?
Pick two or three metrics tied to CSM time and one tied to outcomes. For time: minutes per QBR prep, minutes per renewal brief, tickets handled per CSM per day. For outcomes: gross retention rate, expansion ARR, time to first value for new accounts. Baseline those metrics for a month before rolling out the prompts, then re measure ninety days in. Anything else (number of prompts used, words generated, internal sentiment) is interesting but not the case for keeping the program funded.
Setup

Making these prompts compound

Drop these into a Claude Project loaded with your team's context: your processes, your templates, your past work, and the standards you hold. With those inputs, the prompts above produce outputs far better than running them in a blank Claude chat.

See Claude for Customer Success for the full picture, or have Treetop build the Customer Success Project for your team.

Related

More for Customer Success teams

Turn these prompts into a Customer Success Claude Project
Treetop builds a tuned Claude setup around your accounts, your health model, and your renewal motion, so your CS team runs these plays on real data, not blank prompts. Start with an AI audit or jump straight to implementation.
See Implementation → Start with an Audit
Related

Explore more from Treetop