Airbnb Booking Funnel Analysis

This repository contains the analysis and findings from an investigation into Airbnb's guest-host matching process in Rio de Janeiro. The goal of the project was to identify key opportunities to increase bookings and provide data-driven recommendations for improving conversion rates.

Dataset

The dataset for this analysis can be downloaded here.

Context

The sharing economy is playing an increasingly important role in the tourism industry, with a growing number of individuals acting as either guests or hosts. This trend offers a viable alternative to conventional services and enables tourists to personalize their trips, resulting in enriched experiences. This paper aims to investigate the factors that affect the success rate and speed of bookings in the guest booking process by utilizing funnel analysis, segmentation analysis, and cohort analysis. The study focuses on various factors, including booking channel, host involvement, listing reputation, guest profile, and guest type. The results demonstrate that the success rate of bookings varies depending on the booking channel used (contact me, book it, and instant book), and active engagement and first impression between guests and hosts are two crucial factors in achieving successful bookings.

Keywords: Data Mining, Data Analysis, Two-Sided Marketing Place, Sharing Economy, Funnel Analysis, Cohort Analysis Segmentation Analysis, Airbnb, Decision Making

OKR Model

Objectives:

• Increase the number of bookings in Rio de Janeiro (RDJ).
• Attract new & retain existing customers in the RDJ market.
• Improve the guest-host matching process (speed + convenience).

Key Results:

• Total bookings increases over time.
• Cumulative customer base grows.
• Booking conversion increases over time.

Key Metrics

Conversion funnel:

• Interactions

• Replies

• Acceptances

• Bookings

• Reply Rate (%): # Replies / # Interactions:

• Booking Rate (%): # Bookings / #Interactions based on initial interaction date)

• Acceptance Rate (%): # Accepted Bookings / # Interactions

• Abandonment Rate (%): # Bookings not finalized, but were accepted / # Accepted Bookings

Average time in-between:

• Initial contact -> first reply by host
• Reply -> acceptance
• Acceptance -> booking finalized
• Start -> Finish (complete tunnel)

Total # and % split ot unique:

• First-time customers
• Repeat customers

Possible Key Metric Segmentations:

• Booking channel (Contact Me vs. Book It vs. Instant)
• New vs. existing customer
• Size of guest party
• Length of stay
• Accomodation type (e.g. entire apt/house, private/shared room, etc.)
• Neighborhood location (known vs. unknown)
• Guest origin (= or != host)
• "Completeness" of guest and host profiles (# words)
• Length of initial communication (# words)
• Total # messages exchanged between guest/host

Findings

Based on the analysis of the data, the following conclusions can be drawn:

1. Conversion Rate Differences by Channel: The instant_book channel had the highest conversion rates due to auto-acceptance. The book_it inquiries showed moderate success. The contact_me had the lowest conversion rate, suggesting host response delays or friction in messaging.

2. Supply–Demand Imbalance in January: Booking inquiries surge in January, indicating strong seasonal demand in Rio de Janeiro. However, listing availability and successful bookings do not increase at the same pace, suggesting a supply bottleneck.

3. Segment Performance: Experienced guests (“old users”) and listings with higher review counts had higher conversion rates. Neighborhood performance varied, entire homes converted more frequently.

4. Host Response and Engagement: Longer response times from hosts correlated with lower conversion rates. Listings with more detailed host profiles and descriptions showed higher booking rates.

Recommendations

Based on the findings, the following recommendations can be made:

1. Optimize Instant Book Availability: Prioritize⚡️instant book listing in search & map results. Encourage hosts in high-demand neighborhoods and peak periods such as offering lower service fees in order to reduce booking friction and capture existing demand.

2. Increase Supply Availability During January: Introduce seasonal host nudges and operational initiatives to expand active listings and calendar availability during peak demand periods.

3. Targeted Guest Support for New Users: Provide new user guide to help them shorten the booking time.

4. Further Research and Experiments: Conduct interviews and funnel analysis to better understanding the host concerns with the instant book channel and the reasons for guest to take longer booking time with the contact me channel. Test response-time nudges and host incentives via A/B experiments to improve booking conversion. Leverage price elasticity analysis to optimize pricing and revenue without negatively affecting marketplace demand.

Slides and Appendix

Please refer to the attached slides for a summary of the findings and recommendations. The appendix contains additional content, including code snippets and workbooks used for the analysis.

Thank you for your interest in this A/B test analysis. Should you have any further questions or require additional information, please feel free to reach out to us.