Understanding root cause analysis

What is root cause analysis?

Root Cause Analysis (RCA) is a method to find the underlying reasons for anomalies or issues in a dataset. You examine various data properties and external factors to determine if an anomaly is random or indicates a significant shift. This analysis helps you pinpoint the exact cause behind unexpected changes.

RCA works by breaking down the problem into smaller parts. You look at different data properties, such as user demographics, session durations, or specific events. By examining these, you can identify patterns or correlations that explain the anomaly.

To make RCA more robust, you also consider external factors. These could include holidays, product releases, or marketing campaigns. Such context helps you understand whether an anomaly is due to a broader trend or a specific issue within your dataset.

Steps for conducting root cause analysis

Analyzing anomalous data points

1. Identify and confirm anomalies: Use statistical tools to spot irregularities. For more detailed information, you can refer to Statsig Docs - Analysis with hierarchical ID.

2. Examine data points: Look at specific properties to understand the anomaly. Learn more about the methodologies for this process in Statsig Docs - Data Best Practices.

3. Visualize correlations: Track how different properties relate to the anomaly using visualization tools. You can see examples of this in Statsig Docs - Enterprise Analytics.

Configuring your analysis

• Start with the most queried properties. For guidance, check the Statsig Docs - Setup Checklist.

• Adjust settings to include specific event properties. You can find more on configuring metrics in Statsig Docs - Metric Sources.

• Focus on targeted analysis for deeper insights. For a more comprehensive understanding, consult the Statsig Docs - Running a POC.

Examples of root cause analysis in action

Marketing campaigns

• RCA uncovers issues like a malfunctioning ad campaign when traffic drops. Quick fixes become possible by utilizing tools such as A/B Testing Calculator and understanding the Customer Journey Management.

Product feature launches

• RCA identifies browser-specific bugs when engagement dips. This helps isolate problems affecting specific user groups. Utilizing SDKs and APIs and Integrations can streamline the identification process.

Seasonal trends

• RCA shows how user behavior shifts during holidays. This helps you plan for future changes by analyzing behavioral targeting and implementing strategies using Enterprise Analytics and Conversion Rate Optimization.