Stream analytics events to Amazon S3

Analytics require a Pro or Enterprise plan. Analytics streaming is available only on Enterprise plans.

Stream analytics events to Amazon S3 in near real time to send events to your data warehouse or downstream analytics pipelines without waiting for CSV exports. Organization admins can add multiple destinations, select which event categories to send, and select which deployments to stream from. Analytics streaming uses an inclusion model: no events stream until you select at least one category and one deployment. The same category and deployment selections apply to every destination.

Add a destination

Navigate to the Streaming page of your dashboard.
In the Analytics streaming section, select Configure.
Select Add destination.
Enter an optional label, then provide the Amazon S3 bucket, AWS region, access key ID, and secret access key. Optionally, enter a prefix that Mintlify adds to the beginning of every object key.
Select Add destination.

Prepare Amazon S3 credentials

Create an AWS Identity and Access Management (IAM) user with an access key whose policy allows writing to the target bucket. Mintlify does not support role assumption or temporary credentials that require a session token. At minimum, grant s3:PutObject on the bucket and key prefix that you plan to use. For example:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": ["s3:PutObject"],
      "Resource": "arn:aws:s3:::your-bucket-name/analytics/*"
    }
  ]
}

Use the access key ID and secret access key from that IAM user when adding the destination. The key prefix is optional. Mintlify adds it to the beginning of every object key.

To stop streaming to a destination, open the streaming configuration and select the remove icon next to the destination.

Select streamed categories

Navigate to the Streaming page of your dashboard.
In the Analytics streaming section, select Configure.

Under Streamed categories, select the categories that you want to send.

Category	Examples
Page views	Page and Markdown view events.
Navigation	Navigation item clicks, navigation call-to-action clicks, and version changes.
Search	Search queries, result clicks, search closes, and search comparisons.
Page components	Expandable, accordion, code block, and API playground interactions.
Feedback	Thumbs up, thumbs down, and detailed feedback submissions.
AI Assistant	Assistant conversations, sources, suggestions, feedback, and errors.
Context menu & MCP	Context menu actions, MCP link copies, MCP server installs, and MCP tool calls.

The examples summarize each category. Use the streamed eventType value to identify the exact event.

Select Save changes.

Select streamed deployments

Select which of your organization’s deployments send events. The dashboard lists each deployment by subdomain. If you create a deployment, return to the streaming configuration and select it. New deployments do not automatically start streaming events.

Navigate to the Streaming page of your dashboard.
In the Analytics streaming section, select Configure.
Under Streamed deployments, select the deployments that you want to send events from.
Select Save changes.

No events stream until you select at least one category and one deployment.

Understand streamed data

Mintlify writes newline-delimited JSON (.jsonl) objects to your bucket. Mintlify generates object names from UTC timestamps. If you configure a key prefix, Mintlify writes the objects under that prefix. Each line contains an event envelope with an eventType and payload. The eventType is the event name, such as docs.content.view. The payload contains the analytics event as a JSON object or a JSON-encoded string. If payload is a string, parse it as JSON before loading it into your warehouse.

Example streamed event

{
  "eventType": "docs.content.view",
  "payload": {
    "event_id": "4b91fdbc-4677-4e03-b51b-5f2da41c8654",
    "subdomain": "docs",
    "user_id": "",
    "anon_id": "anon_01JZ8W6QKEJ6ECG1T7QK2S5PZ2",
    "session_id": "session_01JZ8W8CS5JC8T18HXH8ES7Z5M",
    "created_at": "2026-07-22T23:21:41.063Z",
    "event": "docs.content.view",
    "path": "/quickstart",
    "referrer": "https://www.example.com/",
    "user_agent": "Mozilla/5.0",
    "ip": "203.0.113.10",
    "properties": {}
  }
}

The event payload contains the following fields:

Field	Description
`event_id`	UUID that identifies the event. Use it as a deduplication key when loading data.
`subdomain`	Subdomain of the deployment that generated the event.
`user_id`	ID of the authenticated user, when available.
`anon_id`	Anonymous visitor ID, when available.
`session_id`	Visitor or assistant session ID, when available.
`created_at`	ISO 8601 timestamp for when the event occurred.
`event`	Event name. This matches `eventType` in the envelope.
`path`	Documentation path where the event occurred.
`referrer`	Referring URL, when available.
`user_agent`	Browser or client user-agent string.
`ip`	Visitor IP address.
`properties`	Event-specific data as a JSON object or JSON-encoded string.

Configuration changes can take up to one minute to apply.

Streamed events can contain personal data, including IP addresses, user and session identifiers, assistant queries and responses, and feedback comments or contact information. Apply access controls, retention policies, and other data-handling requirements appropriate for your organization.

Get started

Create content

Manage your site

Agent

Assistant

Agent-ready content

Search and SEO

Document APIs

Integrations

Reference

Stream analytics events to Amazon S3

Add a destination

Select streamed categories

Select streamed deployments

Understand streamed data

​Add a destination

​Select streamed categories

​Select streamed deployments

​Understand streamed data

Add a destination

Select streamed categories

Select streamed deployments

Understand streamed data