A guide to Root Cause Analysis problems

5 min read

Hey folks! It is been a long gap since our last post. We apologise to all the people who constantly nudged us to create more informative content. You will get more frequent blog posts from now onwards. Let’s begin

---

RCA cases are intended to test your problem-solving skills. You are given a problem statement, and you have to figure out the root cause of the problem in the interview. Your job as PM requires you to find and tackle unstructured problems. There is no fixed approach to solving an RCA case; each problem would differ. The best method is to ask the right questions and have a thoughtful discussion with the interviewer while solving the problem. 

1. Clarifying Questions: The Problem Statement presented to you in an interview would often be concise (with fewer details) and vague at first appearance. You should always start by asking relevant probing questions to understand more about the business, product and metric which is specified in the problem statement. 

Let's take an example to understand it better.

Statement: We are a SaaS company and have observed an increased churn rate in the last quarter. Identify the issue behind it.

Possible Clarifying Questions: 

• What does our company do, and what is our product?

• How do you define the churn rate for a quarter?

• Where do we operate?

• Who are our competitors? Are they facing similar issues? Is there any new competitor in the market?

2. Structure your approach: In any product, there are several areas you could analyse while searching for a root cause. To ensure you do not miss anything significant, and there is no repetition, always aim to be Mutually Exclusive, Collectively Exhaustive (MECE) in your approach. We will see more about this while solving an example. 

3. Explore all possibilities and find the root cause: After laying out a framework, you have to focus on one segment at a time and explore all possibilities with the interviewer until you identify the root cause. At this step, most people get worried about not finding the solution while going through each segment one by one. Remember to have trust in the process. It will always lead you to the right answer. Keep in mind that the RCA interview is always a collaborative discussion; always keep looking for hints from the interviewer. They will often help you skip segments which are not relevant.

Now, Let’s take an example and solve it together. Before reading the solution to the case, we recommend trying to solve it independently with a friend. 

---

Problem Statement: You are a Product Manager at Linkedin; we have seen a 20% drop in profile views of LinkedIn's high-profile users. Identify the root cause behind this. 

Clarifying Questions

What are high-profile users?

Users who have huge post engagement on LinkedIn.

What do you mean by profile views here? Does this mean how many people visit the profile through LinkedIn and Google searches?

Yes, it means the number of people reaching the profile through Google Search and LinkedIn Search. 

Since when are we facing this issue?

Since Last 6 months

Is this particular to Premium or Non-Premium users? Are there any demographics involved here in the decline of profile views?

It has been seen in both premium and non-premium users. There is no visible trend in demographics for the decline in views.

Why is it a concern to us? Are there any other performance metrics associated with it?

We are concerned that a decrease in profile views could result in less popularity of our platform in the long run.

Makes sense. I will start with analysing direct LinkedIn search first, then jump to the Google search side of things later. Does that sound good?

Let’s limit the analysis to direct LinkedIn searches through the app for now. 

Is there any metric calculation change?

No we haven’t changed any metric calculation manner.

Further Analysis

I have used the LinkedIn app. According to my observation, a user can land on a person's profile through two different user journeys. I will lay down both User journeys and analyse each one of them.

Can there be more user journeys? Like algorithmic change to reduce the number of pop-ups of high-profile users?

No, there is no change in the algorithm

User A

User B

I will analyse the User journey of User A first and then proceed to User B. 

We have data suggesting the ratio of users clicking on names after seeing them on a post hasn't changed. You can focus entirely on User A.

Has there been any change in the size or location of the search bar on the LinkedIn app?

No, there haven't been any changes concerning the search bar on our app. 

Got it. Once selecting the search bar, User A would start typing the name. Are there any bugs that have increased the drop-off rate at this step?

No, according to our data, no changes have been observed here in the drop-off rate. 

Afterwards, that suggestion would start based on - Connection type, Industry relevance, Alumni(Same College/School), and Position of highly active and high-profile persons.

Have we changed anything with the suggestion algorithm here, which is causing different results as compared to before?

Good question. We keep on training the Suggestion algorithm, basis more data we get. But as per our knowledge, this should not create an issue concerning our Profile views.

Got it. According to my analysis, the issue doesn't seem to be internal. Have we changed the way how we calculate Profile views in this period?

No, we haven’t.

I will like to understand more about High Profile Personalities here. Is there any specific gender this has happened with?

No, the drop has been observed for both males and females equally. Though, we have seen a drop mainly for Government and Public sector people. 

Okay, this is an important detail. Was there any political activity/turbulence in the country during this period?

Yes, we had the elections six months back. 

Understood. In my opinion, during the election period, there would be a spike in profile views for Govt. sector, but now it has been normalised to an older range. Is this case for us?

Yes, you are right. There was an increase in the views during this time, but now we are seeing the older trends back in place. Thank you.

---

Hope you liked the example, do tell us in the comments how you approach Root Cause Analysis cases. Also, tell us if you want us to cover any particular RCA questions in the future. Happy to help.