Amplitude integration

What it does

The Amplitude integration lets your agency push product analytics events into a client's Amplitude project and read back the aggregate views they care about. Connect a client's project once and your workflows can record events as users complete signup, place orders, or churn; identify those users with structured properties; alias one identifier to another when a global ID is finally known; and query funnels, event segmentation, and active-user counts to feed dashboards, alerts, or downstream branching logic. A polling trigger watches one or more event types on a fixed interval so workflows can react to changes in segmentation totals without standing up an event collector.

Connect an Amplitude account

1. Open your workspace in TaskJuice and navigate to Connections.
2. Choose Amplitude and click Connect.
3. In a new tab, open the client's Amplitude project, go to Settings, Projects, and select the project you want to connect.
4. Copy the API key and the Secret key from the Project Settings panel.
5. Paste both into TaskJuice. Set the region to us for projects created on amplitude.com or eu for projects on analytics.eu.amplitude.com, then set the Ingestion API Host and Query API Host to the matching defaults (api2.amplitude.com and amplitude.com for US, api.eu.amplitude.com and analytics.eu.amplitude.com for EU).
6. Save the connection.

To rotate the credentials later, generate new keys in Amplitude Project Settings, update the connection in TaskJuice, and the next run will pick them up.

Triggers

• amplitude/event-segmentation-poll polls the Analytics events-segmentation endpoint for a chosen event_type on a fixed interval and emits one activation per cycle, with one record per interval bucket returned since the previous watermark. Amplitude does not push webhooks for product events, so polling is the only way to react to changes in event volume.

Actions

• amplitude/track-event sends a single product event to the HTTP V2 ingestion endpoint with user_id or device_id, event_type, optional event_properties and user_properties, and an insert_id for deduplication.
• amplitude/batch-ingest sends a batch of events through the higher-throughput /batch endpoint, with an optional options object for fields like min_id_length.
• amplitude/identify-user sets or updates user_properties on a user record through the /identify endpoint.
• amplitude/alias-user maps a global user identifier to an Amplitude user_id through the /usermap endpoint, so identifiers minted outside Amplitude can be linked to the canonical record.
• amplitude/query-event-segmentation computes counts, totals, or property aggregates for one event over a date range through the Analytics /events/segmentation endpoint.
• amplitude/get-funnel computes a multi-step conversion funnel through the Dashboard /funnels endpoint, with configurable mode (ordered, unordered, sequential), conversion window, and active-vs-new cohort.
• amplitude/get-active-users returns active or newly created user counts over a date range through the Dashboard /users endpoint.

Known limitations

• Amplitude does not push webhooks for product events, so reacting to user behavior requires polling. The event-segmentation-poll trigger queries the dashboard API on the configured interval and is rate-limited by Amplitude to 360 requests per hour per project; intervals shorter than 300 seconds are rejected at the workflow level.
• The Dashboard query endpoints require both an API key and a Secret key. The ingestion endpoints work with only the API key, but the Secret key is still required when the connection is created because every connection in TaskJuice can drive any action.
• EU-residency projects must be connected with the EU hosts (api.eu.amplitude.com and analytics.eu.amplitude.com). US keys will not authenticate against EU hosts and vice versa.
• Ingestion is asynchronous on Amplitude's side: a 200 OK from /2/httpapi or /batch means the payload was accepted, not that the events have appeared in dashboards. Allow a few minutes for events to surface in event segmentation and funnel queries.
• The query actions return the raw Amplitude response shape (data.series, data.seriesLabels, data.xValues). Workflows that need a flat per-bucket list should pair the action with a Loop node over the interval buckets.