What is Catchbase
Catchbase manages your Apple Ads portfolio. It automatically bids on keywords, splits your budget across campaigns, finds new keywords worth testing, and measures the causal lift your ads actually generate. You set the goals, Catchbase runs against them.
These docs cover everything the platform does. Read the three pages in this section for the overview, or jump to Getting Started if you are here to set up.
What Catchbase does
Four features, all running continuously once you are connected:
-
AI bidding model sets per-keyword bids against the target you choose: cost per install, cost per acquisition on a named event, or return on ad spend. The Bidding Model section walks through how to use it. For the full methodology, see our blog article on reinforcement learning for Apple Ads bidding.
-
Keyword Mining finds new keywords by matching on meaning rather than word overlap. It searches popular and trending keywords across different sources and surfaces relevant, high-intent keywords to add to your Apple Ads campaigns.
-
Budget Allocation recommends how to split a total daily budget across a group of campaigns. It uses the same reinforcement-learning approach as the bidding model, applied at the campaign level instead of the keyword level.
-
Incrementality runs hold-out tests and analyses them with CausalImpact, a Bayesian structural time-series method. It tells you whether an install would have happened anyway and works with the bidding model to bid on incremental keywords.
Reporting, analytics, and the integrations to Apple Ads, Adjust or AppsFlyer, and App Store Connect are how those four features get the data they need.
Key concepts in one minute
- Apple Ads (AA) is Apple's paid search platform, the one Catchbase manages.
- Your MMP (Adjust or AppsFlyer) tells Catchbase which installs came from Apple Ads and what those users did after.
- App Store Connect (ASC) provides app metadata and receives Custom Product Pages from Creative Studio.
- Objective, KPI, target are how a campaign is optimised: what it aims for, what it counts, and the cost or return you will accept.
- Cohort is a group of installs keyed on install date. "Revenue D7" means revenue earned within seven days of install.
- Attribution asks who gets credit for an install. Incrementality asks whether the install would have happened anyway.
Key concepts has the longer versions.
Where to go next
- Setting up for the first time: Getting Started.
- Already running, want to turn on automation: Bidding Model.
- A number in analytics looks off: Metrics glossary.
- Something is broken: Troubleshooting.