Skip to main content

​
Introduction

Amplitude is a product analytics platform built around an event-based data model. Every user interaction is logged as an event with properties, and downstream analyses include event segmentation (totals and unique users per event), funnel conversion analysis, user counts (new and active), and cohort segmentation. Data is scoped to an Amplitude Project — a single workspace inside the Amplitude organization. The connector reads from the Amplitude Dashboard REST API v2 and exposes events, funnels, and user metrics. The list of available metrics is built dynamically from the project’s event catalog, so any event tracked into Amplitude becomes a queryable metric. One data source is created per Amplitude Project; each project has its own API Key and Secret Key. As an analytics connector, session, event, and breakdown data are standardized so widgets can aggregate alongside ad-platform sources without additional configuration. To connect this data source, see How to connect Amplitude to Adriel.

​
Data refresh strategy

​
Architecture data

Architecture covers the Amplitude Project and its event catalog. The catalog is fetched from Amplitude on first use and memoized for 5 minutes per (data source, project) pair, so events newly defined in Amplitude can take up to 5 minutes to appear in the field picker. Two fixed user metrics — New users and Active users — are always available regardless of the event catalog.

​
Reports data

Reports data is fetched on demand. Each dashboard load issues a live request to the relevant Amplitude API endpoint (event segmentation, funnels, or users) for the selected metrics, breakdowns, and date range. Numbers reflect Amplitude’s current ingestion and processing state at request time — there is no nightly snapshot or fixed refresh window.

​
Architecture levels

Organization → Project → Event → User / Cohort An Amplitude Project is the reporting boundary. Events are tracked into the project, and users (and the cohorts derived from them) are the subjects those events describe.

​
Date range limits

Queries spanning more than 6 months are rejected at validation (“We don’t support more than 6 months of data”). Split longer analyses into multiple queries, each within the 6-month window.

​
Metrics

How to read the columnsCode style like active_users is the literal Amplitude field or metric prefix. For per-event metrics, {event} stands for the event display name appended at runtime, and {cohort} is either new or active. Italic text describes how a value is produced when it doesn’t map cleanly to a single Amplitude field or when it’s computed by Adriel.Data type vocabulary: Number, Currency, Percentage, Ratio, Duration, Date, Text, URL, Array, Boolean.
Metrics are generated per event, and per cohortAvailable metrics are built at runtime from the Amplitude event catalog. For each event, the connector emits event, funnel, and frequency metrics both for the full audience and for two cohorts — new users (a rolling 30-day segment defined by Adriel) and active users — so the same event can be compared across cohorts in one widget.

​
Users & sessions

​
Engagement

​
Events & conversions

Funnel and frequency metrics are generated per event, for both the new-user and active-user cohorts.

​
Revenue & monetization

​
Breakdowns

How to read the columnsCode style like country is the literal Amplitude group-by value. Italic text describes a value whose key is configured per data source or produced by Adriel.

​
Geography

​
Device & platform

​
Customer

​
Acquisition

​
Event

​
User property

​
Time

​
Limitations

Based on the current connector, Amplitude has:
  • A 6-month (186-day) cap on the date range for a single query — longer ranges are rejected at validation and must be split into multiple queries (Amplitude Dashboard REST API).
  • Breakdown caps that vary by query mode — event segmentation accepts up to 2 non-date breakdowns; funnel metrics accept at most 1 breakdown and only day date granularity (weekly or monthly raises INVALID_BREAKDOWN_FOR_CHART); user metrics (New users, Active users) accept at most 1 breakdown (Amplitude Dashboard REST API).
  • Unpopulated user counts on non-date breakdowns — New users and Active users return values for date breakdowns and the ungrouped total; breaking them down by a non-date breakdown (for example, country) returns the grouping labels without user counts.
  • No City breakdown — grouping by city is intentionally not offered because Amplitude truncates high-cardinality city results.
  • Truncated high-cardinality results — when Amplitude prunes a large result, a TOO_MUCH_DATA_REQUESTED_INCOMPLETE warning is surfaced and the widget shows the partial data with an explicit caveat.
  • A 5-minute lag on newly defined events — the event catalog is memoized for 5 minutes, so new events appear in the field picker only after the cache refreshes.
  • One data source per Amplitude Project — each project has its own API Key and Secret Key; one project’s credentials cannot read another project’s events.
  • Rate limits from Amplitude — when Amplitude rate-limits a request (HTTP 429), the connector surfaces the API’s response so the query can be retried.

​
API references

​
See also