In recent reliability research, the growing complexity of life-testing experiments demands censoring schemes that balance inferential efficiency with practical feasibility. This study introduces a comprehensive statistical analysis of the Gompertz lifespan model under the Type-II improved adaptive progressive censoring (IAPCS-TII) framework that unifies and extends conventional, hybrid, and adaptive progressive censoring mechanisms. We derive the maximum likelihood estimators for the Gompertz shape and scale parameters and develop asymptotic confidence intervals using both normal and logarithmic transformation methods. Furthermore, a Bayesian inference paradigm is constructed using independent gamma priors, producing posterior estimates and credible intervals under symmetric (squared-error) and asymmetric (linear-exponential) loss functions via Markov chain Monte Carlo simulation with Metropolis–Hastings sampling. The proposed estimators are rigorously compared through extensive Monte Carlo simulations evaluating bias, squared error, average interval length, and coverage probability under various censoring configurations. Moreover, the study explores the optimum censoring configuration that minimizes estimation uncertainty and experimental cost, providing practical guidance for designing efficient life-testing plans. Results demonstrate that Bayesian methods, particularly those based on asymmetric loss, outperform classical inference in terms of estimation precision and interval stability, especially under heavy censoring. A real-life case study on leukemia remission times further validates the practical relevance and superior fitting capability of the Gompertz model using the IAPCS-TII strategy compared to existing schemes.