Dismantling criminal networks, containing epidemics, or mitigating misinformation through selective node removal is a well-studied challenge. Evaluating such efforts requires measuring network strength before and after node removal. A process \(P_1\) is more effective than \(P_2\) if the residual network after removing k nodes via \(P_1\) is weaker than that of \(P_2\) . Most existing metrics rely purely on structural properties (e.g., connectivity, component size) and ignore how practitioners, especially in law enforcement, perceive network robustness. These perceptions often diverge significantly from topology-driven assessments. We propose a novel strength metric that integrates both structural features and human perception. By collecting perceptual feedback via surveys on synthetic and real-world networks, we derive a tunable weight vector capturing perceived importance of connected components. Experiments show our metric aligns more closely with human judgment than traditional methods and improves identification of authoritative nodes for effective dismantling.