Lateral Flow Immunoassays (LFIA) are widely used for on-site diagnosis of various target analytes since they require little to no assisting infrastructures and have simple operations. Regarding AuNPs-based LFIA, the sensitivity can be affected by several factors, including sample quality, sample pretreatment, flow rate, labels, material, etc. Often target analytes tested with LFIA originated in complex biological fluids that could interfere with the assay’s performance. Here, we focus on improving the performance of two types of sample matrices: (1) milk – a protein-rich sample, and (2) saliva – a non-invasive sample type that has a high viscosity. Tween-20 and BSA at different concentrations were employed to enhance the flow rate of the assay and prevent non-specific interactions. Samples were also diluted with a running buffer to further improve the flow rate. We designed a filtration system for the milk matrix to increase the sample concentration. Results showed that sample dilution at a 1:1 ratio was effective in mitigating matrix effects and enhancing flow dynamics for both the milk and saliva matrices. While the milk sample worked better with 0.1% Tween-20, the saliva sample can use either 0.05% or 0.1% Tween-20. The addition of 2% BSA in saliva pretreatment buffer inhibited non-specific reactions, thus enhancing the signal strength of the test line. The pretreatment buffer for saliva matrix showed superior performance compared to the commercial buffer, providing an LOD of 10 ng/ml at 6.12 cP viscosity. Results also highlighted the efficacy of our filtration system, which has a recovery rate of 74.6% of the initial bacterial count. In conclusion, combining Tween-20 and 2% BSA creates a highly effective diluent buffer that can improve the test strip’s performance for complex samples. It is suggested that the experimental buffer be tested with clinical samples, as it contains sufficient ingredients to improve accuracy in the future. This study also provides valuable insights into improving the sensitivity of LFIA when dealing with complex samples.