gtmpodStart
CSMqbr-prep · intermediate

QBR prep in 10 minutes (down from 4 hours)

Last reviewed: 2026-05-23 · saves ~3.5 hr/QBR/run

Our take

QBR prep is the single biggest time-suck for CSMs at companies with > 10 strategic accounts. Most CSMs spend 4 hours assembling product usage data, ticket history, business outcomes, and exec talking points. Claude with the right prompt + a usage dashboard pull does it in 10 minutes. The CSM then spends the saved 3.5 hours on actual customer relationship work.

Tool stack

Vitally

customer-success-platform

Amplitude

product-analytics

Anthropic Claude API

llm-platform

Steps

  1. Set up a Claude Code skill that pulls: Amplitude/Mixpanel usage data, Vitally health record, Jira/Linear ticket history, prior QBR notes.
  2. Skill generates a QBR deck draft with 4 sections: Adoption recap, Wins, Risks, Strategic recommendations.
  3. Output goes to Google Slides via API or Notion page.
  4. CSM spends 10 min editing — particularly the 'Strategic recommendations' section, which AI gets ~60% right.
  5. Send to customer 48 hours before meeting; ask if anything's missing.

Prompts

Generate QBR deck content · Claude Sonnet 4.6
You are a Customer Success Manager preparing a QBR. Inputs:

- Customer name + industry + use case
- Product usage data (last quarter): top features, user count, depth-of-use metric
- Health record from Vitally (current score, trend)
- Support ticket history (volume, severity, themes)
- Prior QBR notes (last 2 sessions)
- Original sale commitments + use cases

Generate 4 sections for the QBR deck:

## ADOPTION RECAP (1 slide, 3 bullets)
Top adoption wins last quarter. Include 1 metric per bullet.

## VALUE DELIVERED (1 slide, 2-3 bullets)
Tie to original commitments. Quote them where possible.

## RISKS + WATCH-OUTS (1 slide, 2-3 bullets)
Adoption gaps, ticket spikes, user churn, engagement drop. Frame neutral.

## STRATEGIC RECOMMENDATIONS (1 slide, 2-3 ideas)
Forward-looking. Tied to customer's stated business goals. Each should be:
- One sentence describing the recommendation
- One sentence on why now
- One sentence on what we can do to help

Output as markdown. Be specific — generic recommendations are forbidden.

Pitfalls

  • AI loves to recommend generic 'expansion opportunities' that aren't aligned with customer goals. CSM must filter.
  • Don't show the customer something they didn't tell you. If a usage metric looks bad, frame it as a discussion not an indictment.
  • Strategic recommendations need CSM judgment. AI should propose; CSM curates to top 2-3.
Last reviewed 2026-05-23. Independent.