Quantifying uncertainty is crucial for enhancing the reliability of geotechnical engineering designs and reducing potential risks. Bayesian inference is commonly used for this purpose, yet the impact of measurement data size on updating results remains underexplored. This study takes the excavation-induced wall deflection as a research demonstration, analyzing the influence of measured data size on Bayesian updates in the context of the TNEC excavation. Results demonstrate that utilizing five or more measured points yields satisfactory predictive outcomes (R2 > 0.90). The variation in MPI values quantitatively reflects decreasing uncertainty with an increasing number of measurement points, while changes in PICP indicate the combined influence of predictive performance and confidence interval width on the probability of measured points falling within the predicted range. This research provides insights into using suitable measurement data sizes for robust parameter calibration and effective predictive performance in Bayesian updating for geotechnical engineering.