# TalkingData MCP > TalkingData connects your AI client directly to a leading big data platform, letting you query complex product and user metrics using natural language. Instead of navigating dashboards or writing SQL, your agent instantly retrieves active user counts, tracks custom events, analyzes cohort retention, and breaks down performance across every marketing channel. ## Overview - **Category:** marketing-automation - **Price:** Free - **Tags:** big-data, product-analytics, user-behavior, event-tracking, attribution ## Description TalkingData connects your AI client straight into a big data platform, letting you ask complex questions about product usage and user behavior using plain English. You skip the dashboard nightmare and don't gotta write SQL queries anymore; your agent just pulls the metrics you need, instantly. When you use this server, your AI client acts like a veteran data analyst who never sleeps. It handles everything from auditing specific user actions to tracking growth trends across every marketing channel—all without taking you off the chat screen. ### Measuring User Activity and Growth You can get real-time stats on how many users are active right now using `get_active_users`. You'll also track new registrations over a specific time frame with `get_new_users`, getting both the total count and detailed signup info. For staying power, you check user retention rates by calling `get_user_retention` to see how many users are coming back. To gauge general product engagement, use `get_session_stats` for usage metrics like average time in session or how often people log in. ### Deep Diving into Behavior and Performance To find out exactly what your app is doing overall, pull a high-level view with `get_app_summary`. You can also see the general setup details of the application itself by running `get_app_info`. Need to know *what* users actually did? You use `list_events` first to grab a complete list of every custom event type being tracked. Then, you drill down with `get_event_data`, analyzing specific behavioral triggers—like when someone 'Add_to_Cart' or 'Profile_Updated'—to understand the user journey in minute detail. ### Understanding Where Users Come From and What Gear They Use To figure out where your growth is coming from, you run `get_channel_data`. This tool breaks down performance by every marketing channel that brings people to your app. For deeper context on who's using the product, use `get_device_stats` to access detailed hardware specs, including OS versions and specific device models, letting you segment users based on their gear. You can also get a general overview of all usage stats through `get_app_summary`. ### How It Works for You Once you subscribe to the TalkingData server, you just plug in your App ID and API Key (the Access Token). Then, you tell your AI client exactly what you want—you don't need to know how complex SQL is. Whether you ask, 'What's our retention rate?' or 'List all tracked events,' the agent handles the query and gives you the clean data back. ## Tools ### get_active_users Gets current statistics on how many users are active right now. ### get_app_info Retrieves general configuration details for the application itself. ### get_app_summary Pulls a high-level overview of overall application performance metrics. ### get_channel_data Gathers specific data points about where users are coming from (acquisition channels). ### get_device_stats Retrieves detailed hardware statistics for the user devices accessing the app. ### get_event_data Analyzes specific, custom-tracked events to understand what users did in detail. ### get_new_users Gets the total count and details of new user registrations over a set period. ### get_session_stats Retrieves usage statistics, like average time spent in sessions or frequency. ### get_user_retention Calculates and reports on how many users return to the app over time (retention data). ### list_events Lists all custom event types that are currently being tracked by the system. ## Prompt Examples **Prompt:** ``` Show me the application performance summary. ``` **Response:** ``` I've retrieved your TalkingData application summary. You currently have 50,000 total users with a 12% growth rate this month. Your most active region is East China. Should I check the retention metrics for your newest cohort? ``` **Prompt:** ``` What is the active user count for today? ``` **Response:** ``` Your active user count for today is 12,500, which is 8% higher than yesterday at this same time. Would you like a breakdown of these users by acquisition channel? ``` **Prompt:** ``` List all tracked events in the app. ``` **Response:** ``` I've listed your tracked events. There are 25 custom events currently active, including 'Registration_Complete', 'Add_to_Cart', and 'Level_Up'. Which event would you like to see detailed analytics for? ``` ## Capabilities ### Measure Current User Activity Retrieve real-time statistics on active users, new signups, and how long users stick around. ### Audit Specific Events Pull detailed analytics for any custom event you've tracked—like 'Add_to_Cart' or 'Profile_Updated'. ### Analyze User Acquisition Sources Break down user performance and growth by the marketing channel they came through. ### Determine Product Engagement Level Get metrics like average session duration and usage frequency to gauge how sticky your product is. ### Profile User Devices Access detailed device hardware stats, including OS version and specific models, for user segmentation. ## Use Cases ### Diagnosing a Drop in Feature Usage The PM notices engagement dropped last week. They ask the agent for `get_event_data` specifically filtered by 'Feature X'. The agent returns data showing that 80% of users stopped triggering this event after using an old mobile OS, leading to a targeted UI fix. ### Evaluating a New Marketing Campaign The Growth Engineer just launched a paid campaign. They ask the agent to run `get_channel_data` for that week and compare it directly against baseline data, isolating the performance increase only from the new channel. ### Understanding Initial User Stickiness The team needs to know if their onboarding flow is working. They use `get_user_retention` on the newest cohort and find out that, while signups are high (`get_new_users`), retention drops sharply after day two, signaling a major UX problem. ### Pre-Mortem Hardware Audit Before releasing an update, the QA team uses `get_device_stats` to check if crashes are spiking on older Android models. The agent quickly pulls device breakdowns, identifying a specific hardware bottleneck that needs fixing. ## Benefits - Stop building reports for basic stats. Use `get_active_users` to get real-time user counts instantly, whether you're checking the morning status or debugging a sudden drop-off. - You can deeply audit specific actions with `get_event_data`. Instead of sifting through raw logs, ask your agent about 'Add_to_Cart' events and get detailed analytics immediately. - Figure out where your users are coming from. `get_channel_data` lets you compare performance across different marketing sources without leaving the chat window. Great for campaign reviews. - Track user loyalty using `get_user_retention`. This tool gives you cohort data, showing exactly how many new signups stick around after 30 days, which is crucial for product health. - Get context on *how* users are viewing your app with `get_device_stats`. You can pull hardware details (OS, model) to see if performance issues cluster on certain phone types. ## How It Works The bottom line is that you talk to your analytics platform instead of logging into its UI. 1. Subscribe to the TalkingData MCP Server and provide your required App ID and API Key. 2. Connect this server to any compatible AI client (Claude, Cursor, etc.). 3. Ask your agent a question like, 'What was our active user count today?' The tool runs the query and returns the structured data directly into the chat. ## Frequently Asked Questions **How do I check if my users are using the app on old hardware with TalkingData MCP Server?** You use the `get_device_stats` tool. This function pulls detailed data like model and OS, letting you segment your user base by device specs to identify performance bottlenecks. **Does TalkingData MCP Server track every single action? What events can I list?** You use the `list_events` tool. It provides a definitive list of all custom event types that are currently being tracked, so you know exactly what data is available for querying. **Can TalkingData MCP Server tell me if my new marketing campaign worked?** Yes. Run `get_channel_data` to see the acquisition sources and compare that output with `get_user_retention` to understand if those new users are sticking around. **What is the difference between using get_app_summary and getting application info?** `get_app_summary` provides key performance indicators (KPIs) like growth rate. `get_app_info`, however, pulls basic configuration data about the app setup itself. **When I use get_app_info, what details does it provide about my TalkingData account?** It pulls core application configuration data. This includes the primary App ID and API Key status needed for all subsequent calls. It confirms that your server connection parameters are correct before you query live metrics. **Are there rate limits if I frequently call get_active_users or get_session_stats?** Yes, the API imposes usage quotas to prevent service overload. We recommend batching related requests where possible and adhering to the documented rate limit schedule for sustained monitoring. **How does TalkingData calculate user retention when I use get_user_retention?** It calculates retention based on defined user cohorts over time. It tracks the percentage of users who return (re-engage) within a specified window after their initial signup date. **Can I filter the results from get_event_data to specific date ranges?** You must include start and end timestamps in the tool call parameters. This allows you to narrow down custom event analytics, making it much easier to audit data for a single campaign period. **How do I find my TalkingData App ID and API Key?** Log in to the [TalkingData portal](https://www.talkingdata.com/), select your application from the management console, and you will find your AppID and API Key (Access Token) in the application settings or 'App Info' section. **Can I query data for a specific date range?** Yes. Tools like `get_active_users` and `get_new_users` accept optional `start_date` and `end_date` parameters (in YYYY-MM-DD format) to retrieve statistics for specific periods. **Is it possible to list all custom events?** Yes! Use the `list_events` tool to retrieve a complete list of all custom behavioral events currently being tracked in your application, along with their unique identifiers.