Who This Helps
You're a Junior Analyst. Your manager just flagged a KPI drop. You need to find the real reason fast and present a fix that sticks. This is for you.
Mini Case
Imagine your team's weekly active users dropped 12% in 7 days. You dig into the data. You find the drop happened after a feature release. But is that the cause? Not yet. You need to diagnose properly.
Do This Now (5 Steps)
- Grab the data – Pull the last 30 days of the KPI. Look for the exact day the drop started.
- Segment the users – Split by region, device, or plan. See if one group caused the drop.
- Check the release log – Match the drop date to a product change. In our case, a new onboarding flow launched that day.
- Run a quick test – Compare behavior before and after the change. Did users complete fewer steps?
- Write one clear recommendation – Example: "Revert the onboarding flow. Test a simpler version next week."
Avoid These Traps
- Blame the first thing you see – The feature release might be innocent. Check other factors like seasonality or a competitor move.
- Skip segmentation – A 12% drop might be 50% in one region and 0% in others. That changes your fix.
- Overcomplicate the analysis – You don't need a fancy model. A simple before-and-after chart works.
- Forget the business context – A drop in sign-ups might be fine if you changed the pricing page. Know the goal.
Your Win by Friday
By Friday, you'll have one page with the root cause, the data behind it, and a clear next step. Your manager will say "Nice work, let's act on this." And you'll feel like a data detective who just solved the case. (Bonus: you can high-five yourself.)