Analytics
The analytics module of Agentic Engine (AE) helps you run all kinds of analyses on data such as user behaviors and properties, so you can make better data-driven decisions. Our analysis models, distilled from industry experience, are industry-leading no-code analysis tools that cover the vast majority of analysis scenarios in the industry. You can save the data conclusions you get from analysis models as reports in a variety of forms and share them with others through dashboards.
In the Basic concepts section, we explain the basic concepts involved in the analytics module and introduce the basic ideas behind the analysis models. Different models suit different analysis scenarios, and the basic concepts help you understand how to choose the right analysis model for your needs.
Events Analysis is the most basic model. It calculates aggregated metrics of specific user behaviors over a period of time, or the trends of those metrics, such as whether DAU is stable from day to day or how much revenue there is today.
Retention Analysis analyzes the retention rate, one of a project's core KPIs. By selecting a first event and a return event, you can quickly get the next-day, 3-day, and 7-day retention of new users. You can also calculate LTV and ROI with Also Show.
Funnel Analysis analyzes the number and percentage of users who complete specified steps in order. You can quickly understand conversion and churn at specific stages, such as the drop-off points in onboarding or level retention data, and spot possible problems.
Interval Analysis analyzes the conversion time between two events that have a causal sequence, such as the median time from registration to first payment, or the distribution of building upgrade times. It can also serve as a complement to Funnel Analysis.
Distribution Analysis divides users into ranges based on their participation. You can divide them by number of times or days of participation, or by the sum of a specific property for each user, such as the cumulative top-up amount, and view the number and percentage of users in each range.
Flows Analysis is generally used for exploratory analysis. With a Sankey diagram, you can see at a glance how users flow in and out before and after key nodes and analyze behavior preferences, such as which activity users join first after logging in each day, or the last core gameplay users engage in before they churn.
Composition Analysis supports cross analysis of two dimensions and lets you compare data across different groups of users. You can quickly understand the profile of a specific user group, design refined operation plans, and create tasks in Engage to push to these users.
Attribution Analysis mainly describes how clicks on multiple in-app resource slots, or the occurrence of multiple events, contribute to a target event. With Attribution Analysis, you can flexibly explore how much each resource slot contributes to the target event, so you can measure the input and output of resource slots and optimize resource placement.
Scenarios - Leaderboard ranks dimensions by metric value and can be used to show Top N lists, rankings, and ranking changes.
Scenarios - Heatmap is a model built specifically for game map analysis needs. With a heatmap, you can analyze how player behaviors are distributed on the map, and compare how different user groups are distributed for the same behavior.
SQL IDE supports custom queries on the data of all projects in the current cluster, including events, users, tags, and cohorts. If the existing models don't fit your analysis scenario, or your metrics involve data from multiple projects, you can calculate directly with SQL statements and display the results with the visualization module.
Based on our experience serving thousands of projects, we've put together best practices for the analysis models that you can refer to alongside your own analysis needs. You can also use the templates feature to quickly create basic analysis metrics.