Root-causing a 20% decline in Facebook ads after a CEO escalation
Company: Meta
Role: Product Manager
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Other
You are a product manager at Meta. Facebook ads have dropped by 20%, and the CEO calls you directly and asks you to fix it. Walk through what you do, from the first minutes after the call to the fix.
```hint Pin down the number before the cause
Before listing causes, settle exactly which ads number fell, what it is being compared with, and whether the drop is real rather than a change in how it is measured.
```
### Clarifying Questions
- Which number dropped: ad revenue, ad impressions served, clicks or click-through rate, or the number of businesses advertising?
- Over what window, and compared with what: the previous day, the previous week, or the same period last year?
- Did the drop happen suddenly or build up gradually, and does it cover Facebook alone or also Instagram and other places the same ads run?
### What a Strong Answer Covers
- A precise definition of the metric that fell, its baseline and the shape of the drop
- Checks that rule a logging, pipeline or definition problem in or out
- A decomposition of the metric into its components, and a cut by segments, that localizes the 20%
- Internal and external hypotheses investigated in a deliberate order, each with the evidence that would confirm or rule it out
- A fix matched to the cause, how you keep the CEO updated, and how a repeat would be caught sooner
### Follow-up Questions
- Impressions are flat but revenue is down 20%. What does that tell you, and where do you look next?
- The cause turns out to be a mobile operating-system privacy change outside Meta's control. What do you do over the next quarter?
- How would you set up monitoring so a drop like this reaches you before it reaches the CEO?
Overview: A metric-drop diagnosis question: Facebook ads have fallen 20% and the CEO asks a product manager to fix it. It tests defining exactly which metric fell, ruling out measurement problems, decomposing and segmenting the drop, ordering internal and external hypotheses, and planning the fix and the communication.