Analyze Account Health

Analyze Account Health

Deep-dive into a B2B account's product usage to prepare for QBRs, assess renewal risk, identify expansion opportunities, or prioritize CS outreach.

Instructions

Step 0: Identify Account & Discover Context

Get the account identifier:

• Company name, org ID, account ID, or group property value
• Ask user if not provided

Search for existing work: Use Amplitude:search to find existing dashboards, charts, or notebooks for this account. If found, ask user if they want fresh analysis or to review existing.

---

Step 1: Quick Health Triage

Use Amplitude:query_dataset to run these queries in parallel:

Usage Trend:

• Event: _active, Metric: uniques, Group by: account property
• Time: Last 60 days, daily interval
• Shows: Activity increasing or decreasing?

Engagement Quality:

• Calculate DAU and MAU for account
• Get DAU/MAU ratio (stickiness)
• Shows: How engaged are active users?

User Momentum:

• Active user count week-over-week
• Shows: Team growing or shrinking?

Classify Health:

• Healthy: Growing MAU, DAU/MAU >40%, positive WoW
• At-Risk: Flat/declining MAU, DAU/MAU 20-40%, negative WoW
• Critical: Steep decline, DAU/MAU 30 days)
• [Count] users ([X]% of licenses)

---

💡 Top Pain Points & Requests

Pain Points

1. [Theme] (X mentions)

• [Concise description]
• Evidence: [Behavioral data] + "[Quote]" - [Source, Date]
• Action: [What to do]

Feature Requests

1. [Theme] (X mentions)

• [What they want]
• Evidence: "[Quote]" - [Source, Date]
• Roadmap status: [On roadmap/Not planned/Considering]

What They Love ❤️

1. [Feature]: "[Quote]"

---

📊 Feature Adoption

High Usage: [Feature] - [X users] (↑Y%) Declining: [Feature] - [X users] (↓Y%) - Investigate Untapped (Upsell): [Premium feature] - Could solve [pain point]

---

🎯 Recommendations

🔥 This Week

1. [Specific action with user/contact name]

📅 This Month

1. [Strategic action with context]

💰 Expansion Opportunities

1. [Upsell signal with evidence]

---

📎 Details

• Analysis Date: [Date]
• Timeframe: [Last X days]
• Confidence: [High/Medium/Low based on data volume]

---

Best Practices

• Always name users - CS needs who to contact, not aggregates
• Connect feedback to behavior - Validate complaints with usage data
• Be specific in recommendations - "Call Sarah about Feature X" not "improve engagement"
• Show trends, not snapshots - Direction matters more than point-in-time
• Flag data gaps - Note low volume, missing properties, or incomplete data
• Prioritize by impact - Focus on issues affecting multiple users or champions

Common Patterns

Churn Risks:

• Champion churned + declining overall usage
• Multiple complaints about same issue + behavioral evidence of friction
• License utilization declining + negative feedback

Expansion Signals:

• Hitting plan limits (users, API, storage)
• Requests for premium features + high engagement
• New users being added + positive feedback

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: amplitude
• Source: amplitude/mcp-marketplace
• License: MIT

Install and usage instructions live in the source repository linked above.