Since quantum coherence was proposed as a resource theory, the research on its quantification and experimental measurement has always attracted widespread attention. At present, studies have proved that collective measurement can improve the measurement accuracy of coherence, but in experiments, the impact of experimental conditions on the performance of collective measurement to estimate quantum coherence is still unclear. In this paper, we numerically simulate the error of collective measurement to estimate coherence under different detection efficiency and resource number, and we compare them with the errors caused by other methods. We also notice that there exists a parameter range where collective measurement method with the detection efficiency of 60% outperforms the tomography method with the detection efficiency of 90%. In addition, with the increase of the number of resources, the performance of collective measurement is almost independent of the quantum state and coherence measure, and the errors are decreasing at the same scale.