Diagnose a KPI Drop: Junior Analyst Root Cause Fix

Who This Helps

You're a Junior Analyst who just saw a KPI drop. Maybe revenue per user slipped 12% in a week. Your boss wants answers by Friday. This guide helps you find the root cause fast and write a recommendation that actually gets used. It's based on the Board Finance & Runway Narrative course, specifically the Runway Trigger Tree mission, which teaches you to trace problems to their source.

Mini Case

Imagine you run a monthly report for a SaaS company. New sign-ups dropped 18% last month. You dig into the data and find that traffic from a key ad channel fell 40% after a budget cut. But wait — conversion rate also dipped 5% because the landing page changed. Now you have two possible causes. Which one do you flag? The course's Scenario Envelope mission helps you weigh both before you decide.

Do This Now (5 Steps)

1. Pull the raw numbers. Get the KPI values for the last 30 days. Don't filter anything yet. Just see the full picture.
2. Segment by channel. Break the KPI into pieces: traffic source, user type, device. Look for the segment that dropped most. In our case, the ad channel traffic fell 40%.
3. Check for timing overlap. Did the drop start right after a change? Maybe a new landing page went live on the same day. Plot the KPI and the change date on a simple timeline.
4. Rank possible causes. List each potential root cause and give it a score from 1 to 3 based on how much it explains the drop. The ad budget cut scores a 3 because it matches the timing and size.
5. Write one clear recommendation. Say: "Restore the ad budget to last month's level and revert the landing page to the old version. Monitor for 7 days." That's it. No fluff.

Avoid These Traps

• Chasing every rabbit. You find three possible causes and try to fix all at once. Pick the biggest one first.
• Ignoring the time lag. A KPI drop might show up 2 days after the real cause. Always check the date of the change.
• Blending segments. Don't average across channels if one channel is tanking. The average hides the problem.
• Forgetting to ask. Sometimes the answer is a quick chat with the marketing team. Don't skip the human step.
• Overcomplicating the recommendation. Your boss wants one action, not a 10-step plan. Keep it simple.
• Not quantifying the impact. Say "restoring the budget could recover 12% of revenue" instead of "it might help."
• Using vague language. Replace "we should consider" with "we recommend." Be direct.
• Skipping the summary. Your boss reads the first line. Make it count.

Your Win by Friday

By Friday, you'll have a one-page analysis that shows the root cause (ad budget cut), the impact (12% revenue drop), and your recommendation (restore budget, revert landing page). Your boss will nod and say "good work." You'll feel like a detective who cracked the case — and honestly, that's a pretty fun feeling.