Root Cause Analysis

Our vision for Amplitude is a proactive one, where meaningful insights are delivered to our Customers, with little to no data digging required. At a minimum, customers should be notified when a spike or dip occurs in their KPIs, but it’s more important to understand why. In the past, identifying why changes occured required random guess work, manual group by-filters, or “eyeballing” charts - all of which could take hours. With Root Cause Analysis (RCA), a new view within Anomaly and Forecast, you can identify the cause of your data anomalies in seconds, at the click of a button. To quote one of our Customers, Hubspot, “RCA literally saved us hours, if not days worth of debugging work and we love it.”

When you run RCA on a detected anomaly, Amplitude analyzes the properties of the anomalous event, and ranks their impact on the change by significance. RCA even factors in external context like region-specific holidays and recent releases in your product.

To give an example, say your active user metrics plummeted, due to a service outage on the west coast. RCA would identify users in Seattle, Portland, and San Francisco as the leading contributors to the change, right away. This level of causal analysis used to be possible for only the most analytically savvy users in Amplitude. Now anyone on the team can do it, which means less time spent digging through data, and more time optimizing your product experience.