Characterizing the dynamics and functional shifts during protein evolution is essential, both for understanding protein evolution and for rationalizing efficient strategies for e.g. enzymes with desired and effective functions.Most proteins organize in families, sets of divergent sequences which share a common ancestor and a have a similar structural fold.We study Aryl sulfatases (ASs), a subfamily of the large and evolutionary old alkaline phosphatase superfamily. Together with their computationally reconstructed putative common ancestors, we expressed and enzymatically characterized these enzymes using micro-fluidics generated high-throughput libraries.We compare the evolvability and robustness of ancestors and four extant ASs which all exhibit catalytic promiscuity towards a range of substrate classes.Sixteen libraries with distinct mutational loads were expressed in individual cells inside micro-droplets.We find that higher mutational loads yielded a longer tail of active variants but fewer total improved variants.The effects of mutational load varied quantitatively between enzymes, meaning in some enzymes a higher, in others a lower mutational load was beneficial, highlighting the importance of testing various mutagenesis regimes.Although ancestors did not display the highest proportion of active variants, they did exhibit the greatest activity increases.We conclude that in ASs the ancestors are not necessarily more promiscuous but the fitness landscape around them is more shallow and thus more amenable to evolutionary optimization.