Avian influenza viruses (AIVs) are a global public health risk; human infection is typically associated with high mortality. While the relationship between several mammalian adaptive mutations and host factors have been described, it is unknown whether additional uncharacterised mutations lead to adaptation. Here, we combine phylogenetic analysis and complementary experimental methods to quantify the impact of novel mutations that emerge at the avian-mammal interface. We constructed phylogenetic trees of mammalian and avian influenza sequences for the polymerase (PA, PB1, PB2) and nucleoprotein (NP) segments and identified potential avian to mammal spillover events. We found >6500 mutations across the polymerase and NP, including known signatures of mammalian adaptation such as PB2 E627K and D701N which occurred independently in mammals 143 and 56 times respectively. We selected 95 mutations which were mostly undescribed and emerged independently multiple times in a range of species and subtypes. Using a minigenome assay in an avian H5N1 backbone to measure the effect of these mutations in human cells we identified PA P28S, NP I425V and G485R as novel mutations leading to polymerase adaptation. In addition, to determine the mechanism of adaptive mutations, we measured polymerase activity in cells lacking a key host factor, ANP32, and cells overexpressing host restriction factors MxA and BTN3A3. Our combined approach revealed novel mammalian adaptive mutations and demonstrated the benefit of combining phylogenetic and molecular approaches in validating novel adaptive mutations.