The COVID-19 pandemic has influenced household housing demand especially with the dramatic increase in working-from-home (WFH) along with reduced mobility because of the restrictions and infection concerns. We provide a comprehensive understanding of changes in housing demand in Singapore where WFH became widespread and many restrictions including the lockdown were implemented after COVID-19. First, a word cloud analysis of the web-scraped data from PropertyGuru, the largest online property listing company in Singapore, suggests a significant change in keywords for housing advertisement. While words related with accessibility or location appeared significantly less in listings, the space and view of housing units were highlighted with words such as the study, unblock view, and renovation. Next, we investigate how the keywords that became popular after the COVID-19 have affected asking housing prices and transaction probabilities of online listings. Finally, we match the keyword information with actual transaction database and analyze whether features reflected in these keywords are indeed associated with changes in the price premium and transaction volume. Our findings capture detailed short-term shifts in housing demand reflected in real time at the online platform and provide important implications to the housing industry and policymakers.