<p>The COVID-19 pandemic has significantly impacted the hotel industry, prompting extensive research on recovery strategies. While studies have explored the influence of hotel facilities, personnel, and tourist behavior on recovery, limited research exists on how spatial characteristics affect hotels' overall recovery performance. This study examines the post-pandemic recovery performance of hotels in Xiamen, China, using a spatially constrained multivariate clustering method (SKATER algorithm) to categorize budget, low-star, and high-star hotels based on their spatial characteristics across three dimensions: business, leisure, and transportation. Additionally, space syntax analysis was employed to evaluate their integration, choice, and mean depth within the urban network. The Kaplan–Meier estimator was used to assess the recovery duration and performance of these hotel subcategories. The findings reveal distinct recovery patterns associated with spatial characteristics: budget hotels demonstrated the strongest recovery capacity, with 94% resuming performance by the end of the year, aligning with their higher integration values (mean 0.821) and lower mean depth values (mean 11.46). Despite their superior environment, high-star hotels experienced slower recovery due to spatial constraints within the urban network and a shrinking target market. Recovery performance among low-star hotels varied by subcategory, with those located in the free trade zone performing the best. Furthermore, the study found that the pandemic significantly weakened traditional seasonal demand fluctuations, reshaping the temporal dynamics of the hotel market. By uncovering the intricate interactions between hotel characteristics, location, accessibility, and recovery performance, this study contributes to the existing literature and provides valuable insights for developing urban resilience and economic recovery strategies.</p>