Metric Overview
| Metric | Description | Exactness | Formula |
|---|---|---|---|
| Sessions | Notification via SDK of app being launched (session_begin). | Exact | Count(filterset AND event_name = _SessionBegin) |
| Revenue_per_install | Total revenue from Purchase events divided by total count of installs. | Exact | Revenue / New_users |
| Active_users | The count of distinct users that had a _sessionbegin or completed any named/integrated event (e.g., register, view, add to cart, purchase) during the specified time range. | HLL estimate | Total_users – New_users |
| Total_users (or Users) | Distinct count of all devices within the filter set (installs, session, events). | HLL estimate | HLL(filterset) |
| Events | Non-distinct count of all completed events (NOT installs, clicks, or sessions). | Exact | Total – (Sessions + Clicks + Installs) |
| Total | Everything within filter set (includes installs, clicks, sessions). | Exact | Count(filterset) |
| Revenue | The sum of the revenue from Purchase events. | Exact | Dimension_Sum(filterset AND event_name IN revenue_events) |
| Events_per_user | Total events (from above) divided by total users (from above). | HLL estimate | Events / Total_users |
| Clicks | The number of clicks during the selected timeframe. | Exact | Count(filterset AND event_name = _Click) |
| New_users | The number of installs during the selected timeframe. | Exact | Count(filterset AND event_name = _Install) |
| RPU | The total revenue/total_users for the user specified date range. | HLL estimate | Revenue / Total_users |
| LTV | By day aggregation of revenue for a defined install cohort. | HLL estimate | Dimension_Sum(filterset AND attribution_date IN cohort) |
| Retained | Essentially the same as Active_users from above, however applied to a defined install cohort date range. | HLL estimate | HLL(filterset AND attribution_date IN cohort) |
| RPI | Same as above. | HLL estimate | HLL(filterset AND attribution_date IN cohort) |
| Average Session Count | The average number of Sessions by a particular user. | HLL estimate | Sessions / Total_users |
| Funnel | Retained users throughout a series of events. | KMV estimate | KMV(filterset AND event_name = X) INTERSECT KMV(filterset AND event_name = Y) INTERSECT … |
NOTE: HLL & KVM Estimates – Visualization of data is based on Theta Sketches (streaming algorithms) to render in real-time and is 97.9% accurate. 100% accuracy available in reporting, postbacks and APIs.
