This week Product School hosted Debapriya Basu current Principal Product Manager at Zillow for an #AskMeAnything session. Debapriya shared her experiences with Artificial Intelligent, gave advice for joining the Product Industry, shared the importance of understanding the customer and more!
Meet Debapriya Basu
Debapriya is a Product Manager that’s exceptional in execution, a very passionate learner, and has a growth mindset. Currently, she’s a Principal Product Manager at Zillow, conceptualizing and incubating AI models that predict user’s state and intent in their real estate journey from browsing for homes to transactions, and designs AI features that make Zillow any buyer, seller, or renter’s trusted personal guide. Prior to her current role, Debapriya was a Senior Product Manager at Microsoft.
The importance of AI
How was your journey into AI?
My journey was non-linear. I was a Computer Science undergrad, spent a few years as a developer, then did my Masters in Computer Science with a focus on Networks and Distributed Systems, joined Microsoft, work in the cloud, moved to Bing, and fell in love with AI. The meta-point is work on what you feel excited about each day. The rest will fall.
How are you addressing ethical and privacy concerns around AI products while designing your strategy and solutions?
I am so glad you are thinking about it. A few ways:
- Obfuscate PII like AD says
- Try to make the training dataset as unbiased as possible.
- AI/ML models used for making decisions or predictions should not be biased concerning protected attributes (latent bias) such as race, gender, and sexuality. It should be aware and counter impacts of interaction and selection bias.
These are starting steps. There are a lot of others. Project Gender Shades is a great initiative. We have a similar initiative at Zillow to have our models be unbiased.
Where are we heading in AI products? What’s your vision for the coming 10 years?
I am glad you are looking far out. I foresee AI becoming much more prevalent than today. I foresee AI being used to solve a lot of problems at scale. I foresee stricter regulations in place to counter the negative side effects of AI. My most favorite dream is the ability to detect illness from gene data at scale using AI. This would lead to better medication for cancer and other terminal diseases. I think and hope that with all the good people on earth, AI will be an effective tool for the positive evolution of mankind.
You might also be interested in: Winning at AI for Product Managers
What are the key hurdles you currently deal with when pushing for conversational AI driven engagement models on Zillow?
The key hurdle as well as the upside is to make AI interpretable. The goal is to have non-technical folks also benefit from this technology. To build trust, my primary goal is to make AI interpretable. So, I often act as the bridge between our scientists and engineers and sales and marketing.
What is your approach in learning and leading a team within a technical and evolving space of AI?
Different kinds of teams work for different AI problems. I try to find the people most suitable to tackle the problem. If it is early incubation, I try to find people who are comfortable with vagueness. If it’s during launch time, I try to find people who are detail oriented. Setting the team’s culture is supreme and also listening, understanding, and caring for your team members go a long way. I have done a webinar at Product School on leading AI incubations. I have a few slides detailing how to build teams and what to keep in mind. I think you will find it useful.
You might also be interested in: A Quick Guide to the Product Management Team
Switching between roles and industries
How can a Senior Product Manager transition to a leadership role? What advice or paths would you define for such a transition?
As a senior product manager, you are also exhibiting leadership. I would urge you to first think “what does leadership mean for you?”, “What kind of leader do you want to be”. Once you have identified the answers, look around to see who exhibits that? follow him/her, study him/her? And then when you feel confident, let your leaders know you are ready for the next level of leadership in your company. Don’t be defined by your level. Be a leader, people recognize. My leadership mantra is to be fearless, be empathetic, help grow the ecosystem. I started doing that even as a junior PM.
As I said, 1) understand what type of leader you want to be 2) grow your skills 3) ask for help 4) help others by sharing what you know 5) tell your leaders, you are ready for the next level.
You might also be interested in: Google Product Manager Reveals Her Leadership and Career Tips
What is your recommendation when it comes to switching between industries?
I have been a Zillow user even when I did not work at Zillow. So, I liked the product, used it, and also thought of ways to improve it. When you switch industries, figure out:
- Why you want to switch
- What do you like about that industry
- How can you make it better
How should I tackle homework assignments for APM positions? Is Zillow hiring any APM this year?
Please check our job pages for open positions. In regards to how to tackle homework assignments, try and understand what the company is asking, use a framework to answer the questions. Try to answer, Why, What, When, How. Also, practice some of these assignments and get feedback from other product people. Great way to learn.
What would be your advice for someone who has worked extensively in technical product development and now wants to move into product management? What should be the learning path?
Think of the customer problem, think about how you will use your technical depth to solve it. Read a lot, discuss. Understand existing product visions. Try to foresee a product in 5 years and think about what else could be done. Good luck. I made that leap and I thoroughly enjoyed it.
Understanding the customer
What are some trends you’re seeing in real estate at Zillow that may not be as visible to others?
Very interesting questions. Some of the trends are:
- People are buying more houses, renting less
- Upgrading to bigger properties
- Moving to remote locations with generous WFH policies instituted during Covid by most tech companies.
Few among many.
How effective are sentiment analysis techniques in gauging customer sentiment?
Yes, sentiment analysis techniques can get erroneous results in the first run. How you retrain the system, add positive and negative reinforcement to the system helps the system to learn. So, post the first run, it is preferable to have humans triage the outcomes and feed into the system. AI systems get better with time and data and reinforcements. All of which we have to give them.
How does one grow in understanding consumer psychology to be a successful PM?
- Discuss with your User researchers and Designers and other PMs
- Study early adopters of products
These are good starting.
You might also be interested in: Customer Communication: What You Are Doing Wrong
How do you make sure that other teams you interface with have a good understanding of the user’s needs? Any tools you recommend for team communication/knowledge transfer?
Engineers must understand product needs. I often set up discussions between engineers, scientists, and designers, researchers. I set up user research programs like Wizard of Oz testing where I invite my engineers to see first hand what users are saying.
At what point does one decide to add the intelligence of AI into an existing product?
Think how AI helps
- Does it help scale?
- Does it help personalize?
- Does it help users do an action quicker?
Please refer to my Product School webinar on Leading AI incubations. You will get pointers.
Advice and Experience
How have you overcome having a small amount of data available?
It is a real problem: my guidance is to build a model on the small volume of data and build UX which allows people to give feedback. Make feedback loops your strength in getting more data to train the model.
What advice would you give to a Product Manager with a poor understanding of product marketing and product strategy?
We all learn, every minute is a learning moment. So I would say someone’s intuition is not yet well defined. Read competitive analysis blogs like Stratechery, try to reverse engineer why a certain strategy was taken, discuss, listen to other product leaders. All of the above will help you grow.
What have you learned when rolling out your map/list view feature? What do your home seekers most love about the experience?
Homeowners like to see visually their potential homes. They like that their experience is personalized based on what previous homes they liked, visited on Zillow. Zillow is their trusted friend providing them all the relevant information in one click or scroll about their houses. The AI personalized list view is a delight for our users. Proud to be part of that team delivering delighter experiences for our users and helping them with a very critical part of their life – home buying.
How do you find or define quality leads at Zillow? What are the criteria you check for?
At Zillow, we consider our leads important. Then we funnel them through filters like how soon the lead needs addressing, what kind of help does the lead need . We use these mechanisms to customize lead responses.
Join us next week for another #AskMeAnything Session for more insights from Product Managers around the world!