Amplitude MCP Server

Amplitude

Amplitude is a product analytics platform for understanding user behavior. This bundle provides access to event data, user properties, funnels, retention, cohorts, and campaign performance metrics.

Entity Hierarchy

• Organization is the top-level account boundary.
• Projects (apps) separate analytics data by product or environment.
• Events are user actions tracked in the product (e.g., "page_view", "purchase").
• Users are identified individuals with properties and event histories.
• Cohorts are saved segments based on behavioral criteria.
• Charts/Reports are saved analytical queries (funnels, retention, event segmentation).

Capabilities

Analytics queries:

• Query event data with filters, grouping, and date ranges
• Run funnel analysis to measure conversion between steps
• Analyze retention curves by cohort or time period
• Segment users by properties or behavior
• View campaign and attribution data

User data:

• Look up individual user profiles and event timelines
• Search users by properties (email, ID, custom attributes)
• View cohort membership and behavioral segments

Configuration:

• List events and their properties tracked in a project
• View user properties and their values
• Access saved charts and dashboards

Workflows

Understand feature adoption: Query event frequency for a specific feature over time. Segment by user properties (plan, region, platform) to identify who uses it most. Analyze retention to see if users come back.

Analyze a funnel: Define a multi-step funnel (e.g., landing page to signup to activation). View conversion rates between each step. Break down by user segment to find where specific cohorts drop off.

Investigate a metric change: Query the relevant event with daily granularity to see when the change occurred. Compare user segments to isolate the cause. Cross-reference with releases or campaigns.

Build a cohort: Define behavioral criteria (e.g., users who performed an action in the last 7 days). Save as a cohort for ongoing tracking. Use the cohort as a filter in other analyses.

Gotchas

• Project scoping: All queries are scoped to a project. Ensure you're querying the right one (production vs staging).
• Event taxonomy: Event names and properties are case-sensitive. List available events first to use exact names.
• Date ranges: Most queries require explicit date ranges. Omitting them may return empty results or time out on large datasets.
• Query latency: Complex queries over large date ranges may take several seconds. Real-time data has a short ingestion delay.
• Sampling: Very large queries may be automatically sampled. Results indicate when sampling is applied.