Bumble Messages Dropped 20%. How Would You Investigate?

My 5-step framework to answer troubleshooting product analytics questions

Photo by Good Faces Agency on Unsplash

Troubleshooting - aka metric investigation - is one of the most common product analytics interview questions. And it’s not just about numbers. It’s about your thinking.

What interviewers look for:

• Can you break down a vague problem into sharp, testable pieces?

• Can you find the data-driven reason something dropped - not just guess?

• Can you recommend something smart and useful based on your findings?

I’ve coached 50+ data analysts and scientists through this exact interview type.

Here’s my 5-step framework to answer the question, “How would you investigate a 20% drop in Bumble messages?”

---

Check out my product analytics interview series:

3 Types of Product Sense Interviews You Need to Master

3 Easy Building Blocks to Master Product Analytics

Don’t Waste Time Applying to FAANG Without Knowing These Metrics

How Would You Evaluate the Health of Facebook Groups?

---

My 5-step framework

1. Lay out the structure of your approach

2. Understand the metric change in detail: Is this metric decline expected? Was it sudden or gradual? Was it caused by seasonality, logging changes or product launches?

3. Build a funnel to isolate the step(s) where the decrease is coming from: What’s the user journey with the app? Where are users falling off? Build a quick funnel: App open → swipe → match → message. Which stage broke?

4. Slice the data: What are key characteristics we can use to segment users to further isolate root causes?

5. Recommend next steps to address the change: What should we do next? How do we fix the issue?

I made a cheatsheet with common causes behind metrics changes, and how to investigate and solve the issue. This can turn your answer into a job offer.

Comment “cheatsheet” or reply to this email with “cheatsheet” and I’ll send it your way.

---

Step 1: Lay out the structure of your approach

You: To answer the question “How would you investigate a 20% drop in Bumble messages?”

I’d like to approach this question in 4 steps:

First, we need to understand the details of why Bumble messaged declined.

Second, if it is concerning, then we’ll build a funnel to understand where the drop-off is coming from.

Third, we can segment users to further pinpoint the metric drop.

Then, we can talk about solutions and next steps.

How does that sound?

Interviewer: Sounds great!

Step 2: Understand the metric change in detail

You: Let’s start with understanding the details of the change. I’d like to check a few things before diving into the funnel.