With the emergence of big data and technological advancements, there is growing interest in incorporating alternative data sources to enhance nowcasting models. This paper investigates the application of alternative data for nowcasting macroeconomic variables, which involves predicting current or near-term economic conditions in real-time or with minimal time lag. Within the information set for forecasting we include a topic attention metric derived from a topic model trained on financial news articles. We implement a proof of concept for nowcasting the US GDP growth using a topic attention metric. Two models, a machine learning-based Long Short-Term Memory model and an econometric-based Mixed-Frequency Data Sampling regression model, are utilized with this alternative data, and other economical indicators, and their performance compared. The results of this study showcase the validity and effectiveness of the topic attention metric in nowcasting macroeconomic variables such as GDP. Overall, this research contributes to the understanding of alternative data’s potential in enhancing nowcasting models and offers insights for applications in finance, policy-making, and business strategy.