Steve Wilcox at asi 2018. The principle of the hybrid model is a data integration which optimises the different strengths of return path data and the TAM panel. But to make the model work we have to sacrifice some of the single-source value in the TAM panel. Is there enough information in the return path data to make up for this sacrifice? And even when we’ve created an audience measurement that is more stable from day to day, is this visible precision hiding a loss in accuracy?