- DROP OFF
- USER DATA
- LOGS
- HUMAN
Short answer
How can I use AI to find where users are dropping off?
Your agent pulls person properties and recent event logs from PostHog to map out user journeys. It identifies the specific steps where users stop interacting. You'll get a clear view of where the friction is happening.
The outcomes.
Where your insights land.
The AI turns raw event data into specific friction points.
DROP OFF
Friction identification
The AI finds the specific event where a user journey breaks. You get a list of the exact steps that lead to abandonment.
USER DATA
Profile analysis
The agent looks at person properties to see if specific cohorts are struggling more than others. You'll know if the issue is universal or segment-specific.
LOGS
Event sequencing
By looking at recent event logs, the AI reconstructs the path a user took before they quit. This shows you the context of the failure.
HUMAN
Decision support
The AI flags the most likely culprits for the drop-off. You decide which product changes to implement based on these findings.
The workflow
What your AI does when the request arrives.
The AI moves from broad user profiles to specific event sequences to find the break in the chain.
Identify the user
The AI pulls the relevant person data to understand who is experiencing the issue.
get_personGather activity logs
It pulls a list of recent actions to see what the user was doing leading up to the drop-off.
list_eventsInspect specific actions
The agent looks at the details of individual events to find errors or unexpected behavior.
get_eventSynthesize the gap
The AI compares the intended flow against the actual event logs to highlight the missing steps.
list_eventsget_event
Try it
Copy these to start.
Use these prompts to begin your investigation.
Starting points
Swap the bracketed text for your actual PostHog identifiers like person IDs or event names.
PostHog Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_vpYk0NiKfoRYdh0mOuVMNfGBNKFlD4K0TVsVAy1i/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — PostHog capabilities are ready to use.
{
"mcpServers": {
"posthog-mcp": {
"url": "https://edge.vinkius.com/vk_preview_vpYk0NiKfoRYdh0mOuVMNfGBNKFlD4K0TVsVAy1i/mcp"
}
}
}Look at the person with ID [ID] and tell me why they stopped using the app after the checkout event.
List the recent events for [User ID] and find where they deviated from the standard onboarding flow.
Analyze the event logs for users who triggered [Event Name] but never reached [Target Event].
Check the properties for [User ID] to see if their subscription tier correlates with the drop-off at [Event Name].
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect PostHog once, then ask.
Just link your account once. Your credentials stay encrypted, and you can immediately start asking your AI about your user data.
Connect PostHog to your AIFAQ
How this task behaves.
- 01
Can the AI change user properties in PostHog?
No. The AI can only read person properties and event logs. It cannot modify your data.
- 02
Can the AI delete events from my logs?
No. This MCP is read-only. Your event history remains untouched.
- 03
Does the AI see all my PostHog data?
The AI only sees the data it needs to answer your specific question based on the tools it calls.
- 04
Can I use this to create new events?
No. There are no tools available to create or trigger new events in PostHog.
- 05
How does the AI know which user to look at?
You must provide a person ID or a specific event context in your prompt so the AI knows which data to pull.
More questions about PostHog? See everything the PostHog Connector can do
Connect PostHog to Claude, Cursor, ChatGPT & more
