Problem Statement
- For apartment rentals, if a customer is interested in renting a place, they will usually visit and compare multiple options, apply for one by paying an application fee to hold it, and then sign the lease after it’s approved.
- As a product manager at a rental startup, I was tasked with increasing the rental application rate to drive final conversions and boost revenue by leveraging the AI leasing assistant via SMS.
Goal & Achievement
- Tour Increase: Boosted tour booking rate by 16%
- Application Increase: Raised application submission rate by 3%
- Revenue Growth: Increased projected annual revenue by $3 million
Methodology derived from this project to drive growth of user behavior
1. Find out the key behavior or step through quantitative and qualitative data
2. Optimize the flow to facilitate the completion of key milestone, such as reducing frictions, and providing more touch points and platforms
3. If there are multiple customer segments, consider creating personalized journeys, or converting customers from one group to another to encourage new user behaviors, such as transitioning non-payers to first-time payers
Product Planning
- Correlation analysis indicated that tours were a key step
There were so many dynamic variants in the sales process, such as availability, price, layout and seasonality. Through the correlation analysis working with data team, I found that property tour had a highest correlation with application: over 90% of customers with submitted applications came visit the apartment before, but not all tours would turn into applications.
- Text, intent analysis indicated a few customers knew exactly what they wanted
Customers were in different stages of house hunting. By analyzing the initial message, I segmented customers with 3 primary groups:
1. ~10% of customers who did their research and knew the exact unit or room they wanted.
2. ~42% of customers who set a few criteria about the apartment, such as ideal move in and price range, which would help narrow down the options.
3. ~20% of customers who were browsing around without specific requirements.
4. The remaining 28% were either fraud, testing, or lacked a phone number.
Solution & Release
With analyses above, I decided to tackle this problem by focusing on these 3 perspectives to improve tour with the main goal of boosting the application.
.png)
- Personalized Customer Journey
Based on the three different customer groups, I designed the personalized journey displayed above to meet their various needs and shorten the tour conversion time.
- Optimized the Tour Flow
A significant drop-off due to qualification questions, including credit scores and cosigners, since the leasing agents didn’t wanna waste time taking customers to tour and then found out they were not qualified to apply. However, some customers felt uncomfortable about sharing personal information like this at the very beginning. To balance the customer experience and workload of leasing agents, I leveraged the AI chat and let it ask about qualifications only when customers showed interest in touring the building instead of in the early conversations.
- Expanded Tour Scheduling Options from Link to Natural Language
1. MVP: Leveraged Calendly's service to quickly test AI conversations with a link and allow customers to self-serve by booking a tour.
2. Iteration: Replaced the external link with an in-house URL to ensure better compatibility and flexibility with internal CRM and other systems, while also allowing customers to access more property information.
3. Improvement: In addition to the self-service link, I led the team in developing a scheduling feature powered by natural language processing to handle cases where customers had questions about the link or when the link was unavailable, accounting for approximately 10% of total cases.
Bits and Bobs
The increase in scheduled tours was not proportional to the rise in applications, which makes sense, as my focus was primarily on improving the booking experience. The key factors in rental decisions, such as housing availability and overall market conditions, are largely driven by supply and demand dynamics.