Product Sense Interview | If you were a PM at Zillow, what will you build and why?

Let's tackle how to answer a product sense interview question. The below case study covers Zillow as the product.

Here is how I will approach this case question:

Clarification Questions

1. Are we talking about specific market?
   Interviewer: USA

2. Any constraints around integration with MLSs or regulatory factors?
   Interviewer: No constraints

Assumptions Made

1. I will assume that I am a PM at Zillow at its current stage, where it has millions of followers

2. No major technical or regulatory constraints

Motivation and Context

User Behaviour:

• Home purchase is the most expensive purchase decision in the individual’s life

• People use Zillow because its get them ability to search homes within their desired constraints

• People browse homes for months or even years before they end up purchasing or leasing one

• Potential home buyers are constantly tracking houses in the areas so that they can end up with good finds

• Zillow is a multi-sided marketplace and agents are also its users. Agents make money when they end up helping the buyer purchase or lease the house

• Home buyers expect transparency on pricing, fees and agent performance metrics. Also demand personalization, virtual tours and simplified checkout experiences.

Recent Trends:

• Interest rates have pushed home affordability down, however longer term trends favour home buying

• Remote work has driven increase in demand in suburban areas

• VR/AR technologies can offer more enriching virtual experience when exploring the house

• AI can disrupt how houses are discovered and come up with interesting insights based on available data

Competition

• Redfin and realtor.com are the major competition

• Opendoor is a competition to Zillow offers product within the Zillow app

Why should Zillow build this product?

Zillow's mission is to "help more people get home — with speed, certainty and ease." Its has a large database of listings and home details to streamline research. In order to advance its mission, Zillow should continue to enhance the user experience and tools so that the customers are able to take purchase decisions with ease and confidence.

Target Segments to consider

This is where it starts getting tricky. Remember there is no right or wrong answer. But do a gut check once you come up with segments.

Before we come up with the segments, let evaluate some of the vectors around which we can segment the users

• People may be interested in specific types of homes → Single Family homes, Townhomes, Condos

• People are interested to live in a certain geography → Downtown vs Suburbs vs Country Side

Now lets look at some user types

• Family Status: Families with Kids vs Individuals

• First time buyer vs Real estate investor vs Renter

• Full time real estate agents vs part time real estate agents (not 100% sure if this is a good user type to consider)

Based on the above vectors and user types, I will prioritize the following segment: Families who are looking to purchase a bigger house in Suburbs

Reasons for Choosing this segment: I believe this segment takes the most involved purchase decision because they need to think about multiple criteria > e.g. school district, commuting time, neighborhoods quality, etc. This segment will have a long conversion cycle and is likely to spend a lot of time on Zillow before they end up purchasing the home.

Note: I am deprioritizing agents as a segment, because this industry is structured in a such a way that balance of power is on the side of the customers (buyer or seller)

Pain Points

Now that we have identified the segment, lets look at some of the pain points for the segment

1. The market (even now) is hot in the desirable areas and thus leads to lot of overbidding for houses

2. The buyer is unsure what is the right value to pay for the house

3. The buyer is unable to compare one house next to the other

4. The buyer has preferences around house specifications and some of these are negotiable while others are not; however, they need to check each and every house individually on the app to come to a decision whether its worth considering

5. The buyer is most times unsure what will be the total cost of ownership including agent fees, legal costs, upgrades, etc.

6. The buyer who is looking to purchase a bigger house is likely to be working in major metros. To fit the budget, they often end up compromising on the commuting distance

7. Photos of the listed houses can be deceptive. You only find out once you have done a tour.

Prioritizations of pain points to be solved

I will prioritize the following pain point:

The buyer is unable to compare one house next to the other

Reasons for prioritizing this pain point: There are multiple reasons to prioritize this pain point:

1. Zillow has large amount of data and insights about the house and locality to solve this problem

2. Solving this problem will help Zillow advance its mission to make it easy and speedy to purchase the house.

Solutions

Note: This is your chance to come up with something outside the box. However, make sure you don’t suggest something that ranks low on practicality.

To solve the pain point of coming up with a fair way of comparing two house, I will evaluate following options

1. Creating an aggregate score for the house and translate that into a recommendation banner

   1. Description: I will come up with a banner on each listing on the home page similar to “best match”, “good match”, “fair match”, etc. To come up with this banner, I will take user inputs on their preferred housing specifications, preferred locations, budget, browsing history, location data, favorite houses, etc. I will come up with a model that will give a new listing a certain score based on the above logic.

2. Build tools to add house to a comparison page, where each house can be compared side by side along with its specific attributes

   1. Explanation/Description: This is similar to the Amazon product comparison that you see on the product detail pages

I will consider option 1 → Creating an aggregate score for the house and translate that into a recommendation banner

Reasons for picking this solution: This solution is more wholistic in nature and does the heavy lifting for the customer. It will solve the problem and help Zillow advance its mission. Additionally, If I am able to provide users the opportunity to give preferences, this will work even better.

Implementation & Monitoring

Note: This may or may not be required. However, to have a good interview rating, I would strongly suggest you propose this section and complete the analysis if the interviewer agrees to proceed.

Implementation

• I will first validate that this problem is big enough > to validate I will look at the NPS data from our customers, our their existing behaviour with our recommended houses

• I will assume the goal is to make customers tour the house

• Once validated, I will work with data scientist and data engineer to come up with model base level assumptions on the weights of each of the above parameter discussed. Depending on how the model performs, we will tune, add/subtract parameters and then agree on an acceptable working model to be tested. The acceptable model would have atleast 70% accuracy on the historic data as measured by people who had certain preferences, ended up touring the house.

• Once built, I will run the A/B test for the model in production with audience for at least two quarters

  • Based on the outcome of the tests, I will further refine the model or scale up the model to more users.

Monitoring

• I will track the model performance month on month and see if the 70% accuracy is still where we should be or should we increase or reduce the threshold

• Metric to track > # of tours scheduled on the property with the ‘best match’ banner within 2 weeks of listing

  • The listing that have best match recommendations have statistically significantly higher number of house tours as compared to listings that don’t.

Summary

We evaluated what will I build as a PM on Zillow. I prioritized the families that are trying to buy a bigger house and decided to solve the pain point of comparing multiple houses. I will build a solution to provide recommendation to the customer based on the level of match for their preference and track the effectiveness of the solution by tracking # of tours scheduled on the property within 2 weeks of listing.

---

Hope you enjoyed the case study. Do you want to practice this case with someone? Join Candidace.fyi to practice and crack your next product management interview.