Metrics glossary
Definitions for every metric surfaced in Catchbase analytics. Each row lists the metric, its formula, its underlying data source, how fresh the value is, and how it behaves across cohorts. Metric names match what you see in the dashboard.
"MMP" means the mobile measurement partner (Adjust or AppsFlyer), the source of installs and revenue. Metrics are grouped using the same sections the dashboard uses on the columns menu: Ad Performance, Conversions, Cost Metrics, Budget, Revenue, ROAS, and the per-event MMP columns.
Three categories: Apple-standard, Auction Insights, and MMP-driven
The metrics split into three groups, and the distinction matters in practice.
Apple-standard metrics are identical across every Catchbase customer. Impressions, taps, tap installs, spend, tap-through rate, conversion rate, cost-per-tap, and cost-per-install come from Apple Ads' Campaign Management reporting. Every customer pulls the same fields, on the same schedule, with the same definitions Apple uses. If two customers compare notes on "CPI", they are comparing the same thing. See References below.
Auction Insights metrics are also standard across customers, but come from a different Apple API (Impression Share) and from Catchbase's own view of keyword state. Average Share of Voice is Apple's; Total Keywords / Prioritized / Deprioritized are counts Catchbase derives from its own bidding state. See the Auction Insights table for the exact definitions, and note Apple's Impression Share API has a 30-day maximum lookback, so SOV values older than 30 days are not available.
MMP-driven metrics depend on how your MMP is configured. Revenue, cohorted revenue (D0, D1, D7, D14, D30), ROAS (total and cohorted), the per-event Count / Revenue / ROAS / CPA columns, and any CPA or ROAS target based on an event all flow from Adjust or AppsFlyer. What gets measured, how events are named, and which events carry revenue are choices made in your MMP setup. Two customers with the same MMP provider can see different custom events, different revenue attribution windows, and different event-to-revenue mappings. Standard cohort windows (D0 / D1 / D7 / D14 / D30) are normalised by Catchbase so the dashboard column shapes are stable; everything beyond that follows your MMP configuration. For the setup side, see Connect Adjust or Connect AppsFlyer; for custom events specifically, see Custom events.
Ad Performance
Conversions
Cost Metrics
Budget
Auction Insights
Auction Insights surface how competitive the auction is and where the bidding model is spending its attention across keywords. Average Share of Voice and Average Bid come from Apple; Total Keywords, Prioritized, and Deprioritized are counts Catchbase computes from its own bidding state. All five appear on the keyword-level Analytics views.
Prioritized plus Deprioritized equals Total Keywords within a given scope.
Revenue
All values come from your MMP (Adjust or AppsFlyer). Cohort maturity rules are explained in full on Data freshness.
Adjust reports the cohorted windows natively. AppsFlyer does not, so Catchbase synthesises them from raw event records using the install time and event time. The user-facing labels and values are the same either way.
ROAS
All ROAS values are computed as MMP revenue divided by Apple Ads spend. Adjust reports its own pre-computed ROAS, but Catchbase ignores it and divides revenue by spend itself so figures are comparable across Adjust and AppsFlyer tenants.
Per-event MMP columns
Each event configured in your MMP produces up to four columns in the analytics tables. The event name is whatever you named it in your MMP (for example, "Purchase", "Signup", "Level Complete"). These columns are tenant-specific by definition.
Per-event columns appear in the column selector only for events that are actively tracked in your MMP and selected for Catchbase. The custom events page covers how an event becomes selectable in the dropdown.
Notes on specific metrics
CPA is event-scoped. There is no single CPA metric. Every CPA is scoped to a named MMP event and appears in the UI as <Event> (CPA). A campaign optimising for a cost-per-acquisition target is optimising against a specific event's CPA.
ROAS is always MMP revenue over Apple Ads spend. Adjust reports its own pre-computed ROAS, but Catchbase ignores that and divides revenue by spend itself. Ratios are computed once at the correct level of aggregation, and figures are comparable between Adjust-connected tenants and AppsFlyer-connected tenants.
Lifetime revenue is never fully mature. Unlike the dated cohorts, there is no fixed window after which lifetime revenue stops growing. Recent install cohorts will always understate true lifetime revenue because later events have not yet happened. See Data freshness.
References
External documentation that defines the underlying metrics and events. Every definition Catchbase uses is consistent with these sources.
Apple Ads (Apple-sourced metrics and Auction Insights)
- Apple Ads product site: https://searchads.apple.com/
- Apple Ads reporting documentation (definitions for impressions, taps, tap installs, spend, and the campaign / ad group / keyword objects): https://developer.apple.com/documentation/apple_search_ads
- Apple Ads Impression Share documentation (the source of Share of Voice; 30-day maximum lookback): https://developer.apple.com/documentation/apple_search_ads
- Apple App Analytics (install and event terminology): https://developer.apple.com/app-store-connect/analytics/
Adjust (MMP-driven metrics)
- Adjust Help Center: https://help.adjust.com/
- Adjust Developer docs (event reporting, revenue fields, cohort exports): https://dev.adjust.com/
AppsFlyer (MMP-driven metrics)
- AppsFlyer Help Center: https://support.appsflyer.com/
- AppsFlyer Developer docs (raw data export and cohort analysis): https://dev.appsflyer.com/
- AppsFlyer cohorts documentation: https://support.appsflyer.com/hc/en-us/articles/207032126
Methodology references (used elsewhere in the docs, relevant here)
- CausalImpact (Brodersen et al. 2015), the method behind the Incrementality section, is not used for these metrics directly but consumes them as its inputs.