Portfolio
Projects
Educational & Professional
Changes in Customer Perception due to Pandemic - Airbnb
The onset of COVID-19 posed an unforeseen threat to the unprecedented growth of Airbnb, being met with major disruptions in form of travel restrictions, government restrictions and social-distancing protocols. Both business and personal travel plans had to be put on hold globally and the sharing economy faced a drop in revenues. This transpired into plummeting occupancy rates on Airbnb from the levels in 2019 (AirDNA, 2020).
The project aims to identify changes in customer attitudes, described through usage of particular words, emotions and topics. All these are accomplished through appropriate data preparation from large text and selective metadata inclusion. The tools utilized for analysis in this research are R and R Studio, leveraging the relevant text mining packages available. SQLite is used as preferred method for faster data ingestion and dataset extraction of large datasets.
Ultimately, aim is to provide such detailed insights into the effect of COVID-19 on customer behaviours, which Airbnb can utilise to make necessary modifications to its services to meet customer needs based on their perception coming out of the pandemic. The significant textual predictors will connect changes in customer attitudes and preferences and provide predictive recommendations.
Airbnb has a publicly available repository of data on “Inside Airbnb” website, which records monthly snapshots of Airbnb datasets. The data for 16 cities from various parts of the world are collected based on the popularity of the cities for tourism. The rationale is that the effects of pandemic, as seen in the most popular cities, will impact Airbnb the most as well and hence, the change in customer attitudes from these cities will be representative of the expected impact to be seen in other regions where Airbnb has active operations.
The overall analysis detects various changes in customer attitudes during the pandemic and their impacts on ratings. But, there are indications that the pre-pandemic attitudes might be returning, based on the predictive algorithms used in the analysis. Temporal projections of the sentiments and topic models show that trends have tended to return to the pre-pandemic levels at the end of 2021, providing basis for deducing that the changes might be temporary. While that is the case, the adaptation of behaviours of customers with the pandemic era might persist even if they don’t translate into ratings. Such cases could be how customers became much less concerned with external facilities around the location of their bookings and the increase in sensitivity towards cleanliness and communication. While Airbnb might not require to substantially change the nature of their services due to the observed transitory changes in customer attitudes, they must persevere to address the attributes that customers have started to increasingly value following the pandemic to not damage the trust that customers have in Airbnb by default since the pre-pandemic era.
Some of the findings and visualisations of the project are as follows:
