# Amplitude MCP MCP > Amplitude connects your AI agent directly to your product analytics data. It lets you analyze user activity, calculate retention curves, and model conversion funnels using natural language queries. Stop building dashboards. Just ask the questions: who is dropping off? Which users are most valuable? How much revenue did this feature generate last quarter? Get answers in real time from all your behavioral metrics without ever leaving your chat window. ## Overview - **Category:** growth-engine - **Price:** Free - **Tags:** user-behavior, product-analytics, retention-metrics, conversion-funnels, event-tracking ## Description Think of it like having a data scientist available 24/7, sitting right next to you. This MCP lets you query deep product behavior data using only plain English. You don't need SQL or Tableau; you just ask your agent about user journeys and get the raw numbers back immediately. For instance, you can track how many users start a process but fail at the payment step, or see exactly which groups of people are most likely to keep using the product month over month. When you connect this through Vinkius, your AI client handles all the heavy lifting, pulling data points—from daily active counts to specific user event streams—and presenting them in a conversational way. It’s real-time insight for product teams and growth managers who need quick answers, not complex reports. ## Tools ### active_users Gets the count of daily, weekly, or monthly active users for a specified date range. ### event_segmentation Queries event counts and unique user numbers based on specific events within a time frame. ### export_events Exports the raw, granular data stream for every event that happened during a specified period. ### get_cohort Requests an export download of behavioral cohorts based on user activity. ### get_funnel Queries the drop-off rate and conversion percentage between a list of defined events. ### get_retention Calculates how many users return to use features over time, given specific acquisition and return events. ### get_user_activity Retrieves the entire sequence of actions for a single user ID, useful for debugging sessions. ### list_cohorts Lists all defined behavioral groups (cohorts) currently tracked in your platform. ### revenue_analysis Pulls the daily financial revenue totals and performance metrics for a given date range. ### search_users Finds specific users using criteria like their ID, email, or device identifier before checking activity. ## Prompt Examples **Prompt:** ``` What's our funnel conversion from 'Landing_Page_View' to 'Sign_Up_Success' for the last 3 days? ``` **Response:** ``` I checked the conversion funnel. Over the last 3 days, 14,200 users triggered 'Landing_Page_View'. Of those, 2,840 completed 'Sign_Up_Success'. This represents an overall conversion rate of 20%, which is slightly above our weekly average. ``` **Prompt:** ``` Show the recent user activity for customer email@example.com ``` **Response:** ``` I've pulled the activity stream for email@example.com. In the last 2 hours, they logged in, triggered 'Payment_Initiated', but then hit an error on 'Payment_Gateway_Failed'. You might want to review the exact property payload. ``` **Prompt:** ``` What is our Daily Active Users (DAU) count since start of the month? ``` **Response:** ``` I fetched the DAU data. Since the 1st of the month, DAU has ranged from 45,000 to 52,000. Yesterday hit a peak at 52,104, reflecting our recent campaign launch. Would you like a breakdown of revenue from this period? ``` ## Capabilities ### Track individual user journeys Retrieve a detailed timeline of every event a specific user triggered, helping diagnose why they abandoned the product. ### Analyze drop-off points in funnels Map out conversion paths step-by-step to pinpoint exactly where users stop moving through your core features. ### Calculate user retention rates Generate curves that show how well specific features keep users engaged over days, weeks, or months. ### Identify adoption trends across groups Query event counts and unique usage numbers over time to see clear patterns in product adoption. ### Monitor daily active users and revenue Instantly pull key metrics, like Monthly Active Users (MAU) or daily revenue totals, into your chat conversation. ## Use Cases ### A user complains about a payment failure. The customer success manager asks their agent to `search_users` by email. They pull up the detailed activity stream using `get_user_activity`, seeing the exact error payload that caused the failed transaction, allowing them to fix the bug immediately. ### Need to prove the value of a new feature. The Product Manager asks for the conversion path from 'Feature X View' to 'Checkout Complete'. The agent runs `get_funnel`, showing that while 50% see the feature, only 15% proceed, proving where marketing needs to adjust its messaging. ### Analyzing seasonal revenue spikes. The Growth Manager asks for DAU and daily revenue trends since the start of the year. The agent executes `active_users` combined with `revenue_analysis`, confirming that campaign launches consistently boost both user count and cash flow. ### Understanding long-term product stickiness. The team needs to know if new users are sticking around. They ask for a cohort analysis, running `get_cohort` to see which groups of users acquired last month are still active today. ## Benefits - Stop guessing why users leave. You can request a `get_funnel` analysis to pinpoint the exact step where drop-off is highest, turning assumptions into data points. - Debug single accounts instantly by running `get_user_activity`. Instead of digging through logs, your agent pulls up an individual user's complete event stream for review. - Measure long-term product health with `get_retention`. This tool generates the day-over-day curves showing if new features actually keep people coming back. - Track business impact immediately. Use `revenue_analysis` to pull daily revenue data alongside your user counts, giving a full picture of value per cohort. - Cross-reference findings easily. You can run `list_cohorts` and then use the resulting groups to analyze specific event segmentation, connecting behavior to outcomes. ## How It Works The bottom line is: you get instant, conversational access to complex product analytics without building a single dashboard. 1. Subscribe to this MCP and provide your Amplitude API Key and Secret Key. 2. Connect it to your preferred AI client (like Cursor or Claude). 3. Ask a product question naturally. Your agent queries the data, then gives you the direct answer. ## Frequently Asked Questions **How do I use Amplitude MCP to calculate retention rates?** You run the `get_retention` tool by specifying your start event (acquisition), return event, and date range. This gives you a curve showing how long users stick around after adopting that feature. **Can I check raw user activity using Amplitude MCP?** Yes. Use `get_user_activity` by providing the user's ID. It pulls up the entire event stream, letting you see every click and action they took in detail. **How does Amplitude MCP help with funnels?** You use `get_funnel`, passing comma-separated event names. The agent calculates the conversion percentage at each step, showing exactly where users are dropping off. **What is the best way to check active users? (Amplitude MCP)** Use the `active_users` tool by providing a date range. This quickly gives you reliable daily, weekly, or monthly counts for your key metrics. **How do I authenticate Amplitude MCP with my account keys?** You need to provide your specific Project API Key and Secret Key from Amplitude. Input these credentials into Vinkius; this establishes the connection needed for all subsequent queries. **What is the best way to pull raw event data using the export_events tool?** The `export_events` tool lets you grab massive amounts of raw data over a date range. Remember that API rate limits apply, so it's best practice to narrow your initial time window. **How does Amplitude MCP help me manage or analyze behavioral cohorts?** Use the `list_cohorts` tool to view all existing user groups. This lets you understand how users are clustered by behavior, which is key before deep analysis. **What happens if my Amplitude MCP query fails due to complexity or volume?** If a query fails, it usually means the scope was too large for one request. Try simplifying your date range or breaking down the query into smaller segments first.