Preset metrics cover common behaviors, but product-specific questions—such as which checkout steps drive abandonment or which app versions have the highest retention—require instrumentation you define. Custom analysis lets you instrument, collect, and analyze app events beyond preset metrics, using a client-side instrumentation model backed by an Online Analytical Processing (OLAP) engine for real-time, multidimensional visibility.
How it works
Custom analysis uses a client-side instrumentation model with four components:
Event metadata configuration management
Custom analytical models
Custom analysis patterns
Frontend custom analysis reports
These components work as follows:
Event metadata configuration management: Defines the metrics your app sends for real-time analysis. This metadata determines the tables created in the OLAP database and the dimensions available for queries.
Custom analytical models and Custom analysis patterns: Models represent specific analysis types, such as event analysis and funnel analysis. Each model enforces its own query structure. Patterns are reusable configurations that capture a specific combination of model, query conditions, and dimensions. Save patterns to reproduce analyses you run frequently.
Frontend custom analysis reports: Reports generated from a model and a set of query conditions. Aggregate reports into a custom dashboard to monitor the metrics that matter to your team.
Analysis types
Two analysis types are supported:
Custom event analysis: Track and break down events you instrument yourself. Use the mPaaS OLAP engine to filter, group, and compare events by any dimension you define.
Behavior analysis: Monitor standard engagement metrics across your user base: retention rate, user and device access trends, device distribution, version distribution, hot spot pages, and access time distribution.
Get started
To run your first custom analysis:
Define your event metadata to specify the metrics and dimensions your app will send.
Choose an analytical model that matches the question you want to answer—for example, event analysis for a breakdown by property, or funnel analysis for a step-by-step conversion view.
Configure query conditions and save the configuration as an analysis pattern if you plan to reuse it.
Generate a report and add it to a custom dashboard.