A Claude Skill that takes a customer account and generates the content for a quarterly business review deck: usage trends, top wins of the quarter, open risks, success-plan progress, and proposed expansion paths. CSMs go from a blank Google Slides template to a 70%-done draft in five minutes.

What you’ll need

Claude Code or Claude.ai with the Skill installed
Salesforce read access (Account, Opportunity, Case objects)
Gong API key with read access to call recordings
Your QBR deck template (Google Slides or PowerPoint) with named placeholder slots
A success-plan field or doc per account (Notion, Gainsight, or a Salesforce custom object)

Setup

Install the Skill. Drop qbr-prep.skill into ~/.claude/skills/. The Skill defines a single command, prep_qbr(account_id, quarter), plus three internal helpers for Salesforce, Gong, and slide assembly.
Wire credentials. Set SFDC_TOKEN, GONG_API_KEY, and either GOOGLE_SLIDES_TOKEN or a path to a local PPTX template in your Skill environment. The Skill reads them on init.
Map slide placeholders. Edit templates/qbr.json to map placeholder names ({{usage_trend}}, {{top_wins}}, {{risk_summary}}) to the Skill’s output sections. This is a one-time setup per template.
Run for one account. prep_qbr(account_id="0014x...XYZ", quarter="Q1-2026"). The Skill produces a populated draft deck and a one-page exec summary. Review, edit, send.

How it works

The Skill pulls four data streams in parallel: Salesforce account history (renewals, expansions, contract value over time), Salesforce cases (volume, severity, time-to-resolve), Gong calls tagged to the account from the last 90 days, and the success-plan doc.

It then runs three Claude passes. The first pass extracts themes from Gong call transcripts: what executives said about value, what blockers were raised, and which competitors were mentioned. The second pass cross-references usage data with success-plan goals to compute a progress-against-plan summary. The third pass writes the actual slide content in your team’s voice — neutral, data-led, no superlatives.

CSMs always edit before sending. The Skill is a draft engine, not a publisher.

Watch-outs

Gong call coverage. If the CSM hasn’t logged calls or invited Gong to recurring syncs, the Skill produces thin themes. Audit Gong coverage per account before relying on this.
Success-plan freshness. If success plans aren’t updated, the progress-against-plan section misleads. Better to leave the section blank than fill it from stale data — set the Skill to flag missing inputs rather than invent.
Tone drift across CSMs. Each CSM has a voice. Run the Skill output through a “voice match” pass against three sample QBRs from that CSM if tone matters.

Stack

Salesforce — account, opportunity, and case history
Gong — call themes, executive quotes, competitor mentions
Claude — synthesis and slide content generation