Atlan Usage Analytics

SQL query library and Claude Code skills for Atlan usage analytics. Works with any Snowflake instance where the Atlan Lakehouse is configured (PAGES, TRACKS, USERS tables).

Prerequisites

• Claude Code installed
• Snowflake account with access to your usage analytics data
• Snowflake MCP server configured in Claude Code

Setup

1. Clone the repo

git clone https://github.com/gregmartell-atlan/atlan-usage-analytics.git
cd atlan-usage-analytics

2. Configure Snowflake MCP

Add to your Claude Code MCP settings (~/.claude/mcp.json or via Claude Code settings):

{
  "mcpServers": {
    "snowflake": {
      "command": "npx",
      "args": ["-y", "@anthropic/snowflake-mcp-server"],
      "env": {
        "SNOWFLAKE_ACCOUNT": "<your-account>",
        "SNOWFLAKE_USER": "<your-user>",
        "SNOWFLAKE_PASSWORD": "<your-password>",
        "SNOWFLAKE_WAREHOUSE": "<your-warehouse>"
      }
    }
  }
}

3. Configure your database

Edit CLAUDE.md and update the Configuration table at the top:

Variable	What to set
DATABASE	Your Snowflake database name (e.g., MY_DATABASE)
SCHEMA	Your schema containing PAGES, TRACKS, USERS (e.g., usage_analytics)
SNOWFLAKE_MCP_TOOL	Your MCP tool name (default: mcp__snowflake__run_snowflake_query)

Or simply run /setup in Claude Code and it will walk you through it.

4. Verify data access

Open Claude Code in this directory and run:

/analyze table profiler

This executes table_profiler.sql to verify connectivity and data shape.

Skills (Slash Commands)

Skill	Description	Example
/setup	Configure Snowflake connection — run this first	/setup
/analyze	General analytics — finds the right query or writes custom SQL	/analyze MAU for acme.atlan.com
/discover	Browse available events, pages, domains	/discover events
/users	Active users — MAU/DAU/WAU, stickiness, power users	/users trends acme.atlan.com
/features	Feature adoption — pages, events, matrix, quadrant	/features quadrant acme.atlan.com
/engagement	Session depth, actions per session, daily patterns	/engagement acme.atlan.com
/retention	Cohorts, churn, reactivation, funnels	/retention cohort acme.atlan.com
/health	Customer health score, license, roles, alerts	/health acme.atlan.com
/qbr	QBR data pack, multi-customer compare, risk alerts	/qbr acme.atlan.com

Data Architecture

Table	What	Volume	Key columns
PAGES	Page views	~3-4K/month	user_id, domain, name, timestamp
TRACKS	Events	~80K/month (12K after noise filter)	user_id, event_text, timestamp
USERS	Identify snapshots	~333 rows	id, email, role, domain (~2% match rate)

Key: PAGES.domain is the primary domain source. user_id is the primary identity key. USERS provides optional email/role enrichment via LEFT JOIN.

Note: System-generated workflow events (100K+/month) are excluded by default. Use the "Include workflows?" option in any skill to include them.

Query Categories

00 - Schema Profile

Query	Description
table_profiler	Data availability check: row counts, column fill rates, user overlap
discover_events	Event catalog — all tracked events with usage stats
discover_pages	Page catalog — all page names with view counts
discover_domains	Domain catalog — all customer domains with activity levels

01 - Active Users

Query	Description
mau_by_domain	Monthly active users per customer domain with MoM delta
dau_by_domain	Daily active users per domain
wau_by_domain	Weekly active users per domain
mau_dau_ratio	DAU/MAU stickiness ratio (>0.3 = strong, <0.1 = episodic)
user_roster_by_domain	Full user list with last activity, total events, status

02 - Feature Adoption

Query	Description
top_pages_by_domain	Most visited Atlan pages per domain
top_events_by_domain	Most frequent tracked events per domain
feature_adoption_matrix	Feature-by-user boolean matrix per month
feature_trend_weekly	Week-over-week feature usage trends
feature_engagement_quadrant	Feature reach vs depth analysis (engagement matrix)
connector_usage	Data source/connector interaction patterns

03 - Engagement Depth

Query	Description
session_duration	Session length analysis per domain (monthly, time-gap derived)
session_duration_daily	Daily session duration (avg/median in seconds)
power_users	Top users by composite activity score
actions_per_session	Average events/pages per session
engagement_tiers	Power/Heavy/Light/Dormant user segmentation
daily_engagement_matrix	Daily event counts bucketed into engagement tiers
avg_pageviews_per_user_daily	Average page views per active user per day

04 - Retention

Query	Description
monthly_retention_cohort	Cohort retention matrix
activation_funnel	New user activation rates (1d/7d/14d/30d)
churned_users	Users lost month-over-month with context
reactivated_users	Users who returned after 30+ day gap
daily_retention_session_to_pageview	Day-N retention: activity to pageview (14-day window)
daily_retention_session_to_search	Day-N retention: activity to search/AI action
daily_retention_session_to_session	Day-N retention: activity to return activity
retention_rate_aggregate	Aggregate retention rate: activity to pageview within 7 days
funnel_session_to_pageview	Multi-step funnel with governance split

05 - Customer Health

Query	Description
customer_health_scorecard	Composite 0-100 health score per domain
domain_summary_snapshot	One-row summary per domain, latest month
license_utilization	Active vs total users by domain/role
role_distribution	Role and job_role breakdown per domain

06 - CS Review

Query	Description
qbr_deck_data	All-in-one QBR data pull for a single customer
multi_customer_comparison	Side-by-side domain comparison
trending_alert	Proactive risk flags (declining MAU, low stickiness)

Parameters

All SQL files use {{placeholders}}. Claude Code substitutes these automatically when using skills. For manual execution:

• {{DATABASE}} - Your Snowflake database name
• {{SCHEMA}} - Your schema name
• {{DOMAIN}} - e.g., 'acme.atlan.com'
• {{START_DATE}} - e.g., '2025-01-01'
• {{END_DATE}} - e.g., '2025-12-31'
• {{MONTHS_BACK}} - e.g., 6
• {{RETENTION_DAYS}} - e.g., 14 (daily retention window)

Directory Layout

CLAUDE.md                      - Configuration + context (edit this first)
sql/00_schema_profile/         - Data availability checks
sql/_shared/                   - Reusable CTE snippets
sql/01_active_users/           - MAU, DAU, WAU, stickiness
sql/02_feature_adoption/       - Pages, events, feature matrix
sql/03_engagement_depth/       - Sessions, power users, engagement tiers
sql/04_retention/              - Cohorts, churn, activation, funnels
sql/05_customer_health/        - Composite scores, domain summaries
sql/06_cs_review/              - QBR prep, comparison, alerts
.claude/skills/                - Claude Code skill definitions
models/                        - Semantic model (reference only)