Since early 2024, a multistate outbreak of highly pathogenic avian influenza H5N1 has been affecting dairy cattle in the USA. The influenza viral RNA concentrations in milk make it an ideal matrix for surveillance purposes. However, viral RNA detection in multi-component fluids such as milk can be complex, and optimization of influenza detection methods is thus required. Raw bulk tank milk and mastitis milk samples were artificially contaminated with an avian influenza strain and subjected to five extraction methods. HCoV-229E and synthetic RNA were included as exogenous internal process controls. Given the high viral load usually observed in individual raw milk samples, four out of five tested methods would enable influenza detection in milk with normal texture, over a time window of at least 2 weeks post-onset of clinical signs. Nevertheless, sample dilution 1:3 in molecular transport medium prior to RNA extraction provided the best results for dilution of inhibitory substances and a good recovery rate of influenza RNA, that reached 12.5 ± 1.2% and 10.4 ± 3.8% in two independent experiments in bulk milk and 11.2 ± 3.6% and 10.0 ± 2.9% on two cohorts of mastitis milk samples. We have also shown compatibility of an influenza RT-qPCR system with synthetic RNA detection for simultaneous validation of the RNA extraction and RT-qPCR processes.