Amplitude

Connect & Ingest data from Amplitude

Amplitude is a product analytics platform that helps teams understand user behavior, track engagement, and measure retention. The Sling Amplitude connector extracts data from the Amplitude REST API, supporting behavioral cohorts, chart annotations, event metadata, time-series metrics like active users and session length, and raw event export.

CLI Pro Required: APIs require a CLI Pro token or Platform Plan.

Setup

The following credentials and inputs are accepted:

Secrets:

• api_key (required) -> Your Amplitude project API Key

• secret_key (required) -> Your Amplitude project Secret Key

Inputs (optional):

• anchor_date -> Starting date for first sync of time-series endpoints (default: 1 year ago). Format: YYYY-MM-DD

• base_url -> Override the base URL (defaults to https://amplitude.com)

Getting Your Credentials

1. Log in to your Amplitude account

2. Go to Settings (gear icon) → Organization settings → Projects

3. Click on your project name

4. You will see the API Key and Secret Key on the project settings page

5. Copy both values

Important: The API Key and Secret Key are used together for HTTP Basic authentication. Keep your Secret Key secure and never share it publicly.

Using sling conns

sling conns set AMPLITUDE type=api spec=amplitude secrets='{ api_key: "your_api_key", secret_key: "your_secret_key" }'

Environment Variable

export AMPLITUDE='{ type: api, spec: amplitude, secrets: { api_key: "your_api_key", secret_key: "your_secret_key" } }'

Sling Env File YAML

See here to learn more about the sling env.yaml file.

connections:
  AMPLITUDE:
    type: api
    spec: amplitude
    secrets:
      api_key: "your_api_key"
      secret_key: "your_secret_key"

With anchor date for historical data:

connections:
  AMPLITUDE:
    type: api
    spec: amplitude
    secrets:
      api_key: "your_api_key"
      secret_key: "your_secret_key"
    inputs:
      anchor_date: "2024-01-01"

Replication

Here's an example replication configuration to sync Amplitude data to a PostgreSQL database:

source: AMPLITUDE
target: MY_POSTGRES

defaults:
  mode: full-refresh
  object: amplitude.{stream_name}

streams:
  # Metadata endpoints
  cohorts:
  annotations:
  events_list:

  # Time-series metrics (incremental)
  active_users:
    mode: incremental
  average_session_length:
    mode: incremental

  # Raw event export (incremental)
  export:
    mode: incremental
    primary_key: [uuid]
    update_key: server_upload_time

Full refresh for all endpoints:

source: AMPLITUDE
target: MY_POSTGRES

defaults:
  mode: full-refresh
  object: amplitude.{stream_name}

streams:
  cohorts:
  annotations:
  events_list:
  active_users:
  average_session_length:
  export:

Endpoints

Metadata & Catalog

Endpoint	Description	Incremental
cohorts	Behavioral cohort definitions	No
annotations	Chart annotations	No
events_list	Active event types with current week totals	No

Time-Series Metrics

Endpoint	Description	Incremental
active_users	Active and new user counts by day	Yes
average_session_length	Average session length in seconds by day	Yes

Raw Event Export

Endpoint	Description	Incremental
export	Raw event data (all event properties, user properties, device info)	Yes

The export endpoint returns raw event data in zipped JSON-lines format. Each record contains the full event payload including event_type, event_properties, user_properties, device_id, session_id, timestamps, and more. Data has a 2-hour latency from the time events are received by Amplitude. Days with no events return no records (404 responses are handled gracefully).

To discover available endpoints:

sling conns discover AMPLITUDE

Incremental Sync

The active_users, average_session_length, and export endpoints support incremental sync using date-based iteration:

• First run: Fetches data from the anchor_date (default: 1 year ago) up to yesterday (or 2 days ago for export, due to data latency)

• Subsequent runs: Only fetches data from the last synced date onward

For active_users and average_session_length, each day produces one record containing the time-series data for that date, with a date column added for tracking. For export, each day can produce many records — one per raw event — with uuid as the primary key and server_upload_time as the update key.

To customize the starting date:

connections:
  AMPLITUDE:
    type: api
    spec: amplitude
    secrets:
      api_key: "..."
      secret_key: "..."
    inputs:
      anchor_date: "2024-06-01"

Backfill

To backfill a specific date range, use the range source option:

source: AMPLITUDE
target: MY_POSTGRES

defaults:
  object: amplitude.{stream_name}

streams:
  active_users:
    mode: incremental
    source_options:
      range: "2024-01-01,2024-06-30"
  export:
    mode: incremental
    primary_key: [uuid]
    update_key: server_upload_time
    source_options:
      range: "2024-01-01,2024-06-30"

Rate Limiting

The Amplitude API enforces rate limits that vary by plan:

• Free/Starter: Limited API access

• Growth/Enterprise: Higher rate limits

The connector automatically:

• Uses conservative rate limiting (3–5 requests/second for most endpoints, 1 request/second for export)

• Processes time-series endpoints with concurrency of 3 (export uses concurrency of 1 due to larger payloads)

• Retries with exponential backoff on 429 (rate limit) responses

• Allows up to 5 retry attempts with a base backoff of 5 seconds

If you are facing issues connecting, please reach out to us at [email protected], on discord or open a Github Issue here.