Has anyone seen a great product for “root cause analysis”? (understanding why a metric has suddenly fallen 5%, or why variant A of an experiment is outperforming variant B etc).
I understand that it is a hard problem to solve with a lot of variables but I want to know what are the available state of art tools. Thank you!
It’s called a data analyst
But seriously, Amplitude just released their beta feature called Root Cause Analysis. It works only for anomalies now (when there is an unexpected spike or a data gap). It scans the anomaly across all other available events and properties to help you locate the pattern, or a cause.
https://sisudata.com/ also claims to do this but I have not used it.
Does anyone have any feedback on sisu?
I have used SISU and there were a few challenges we faced:
- It requires careful thought process on input variables. Similar to what Ankit mentioned, you need to rely on a bit of intuition on variables that could’ve possibly impacted the KPIs
- Redundant output. Making it up let’s say - CVR was impacted due to steep CVR drop among 1. Males x Texas and 2. Males
- Output may not always be actionable
This topic was automatically closed after 180 days. New replies are no longer allowed.