A fundamental challenge to media planners in using television advertis
ing is to choose a daypart (time of day) combination that will achieve
efficient delivery of an audience. The author addresses audience accu
mulation by daypart allocation with a household-level exposure model u
sing viewing records from single-source data. By retaining the behavio
ral assumption of a popular beta binomial distribution model, exposure
by each household within each daypart during each week is assumed to
follow a binomial process with a distinct exposure probability that is
inferred from the household's past exposure history. The model captur
es exposure heterogeneity nonparametrically by household-specific expo
sure probabilities rather than fitting a parametric beta function and
automatically accounts for audience duplication across dayparts. This
approach is made possible by the availability of large-scale household
-level exposure records obtained from audimeters and the increasing co
mputation power of personal computers. An example of spot allocation t
o dayparts shows hour media planners can use the model to obtain insig
hts into audience accumulation.