Brandometer, Airbnb's natural language understanding technique, quantifies brand perception from social media data. It measures brand perception by calculating the relatedness of words to "Airbnb" using word embeddings and cosine similarity. The data is cleaned and processed to mitigate the impact of noise and sparsity. Various word embedding models are explored, with DeBERTa performing best with small datasets. To stabilize the scores, averaging and ranking techniques are employed. The calibrated scores are assessed using metrics like Average Variance Across Different Periods (AVADP). Brandometer has proven useful for industry analysis, identifying emerging perceptions, and tracking the impact of campaigns. Future improvements include better content segmentation and enhanced data foundation. By capturing brand perceptions from social media, Brandometer provides valuable insights for enhancing customer experiences.