Some analytic providers use weekly dummy variables to capture seasonality in sales models. Theoretically, estimating weekly dummies jointly across all category SKUs with similar seasonal patterns can be one of the best methods of estimating seasonality. In practice, however, dummies are fit one SKU at a time. Our analysis shows that dummies in single SKU or brand models pick up other effects like price and promotions. Further, these dummies over-parameterize the model and contribute to poor holdout sample performance.