Consumer price indexes (CPIs) are commonly compiled at the higher (weighted) level using Laspeyres-type arithmetic averages. This paper questions the suitability of such formulas and considers two counterpart alternatives that use geometric averaging, the Geometric Young and the (price-updated) Geometric Lowe. The paper provides a formal decomposition and understanding of the differences between the two. Empirical results are provided using United States CPI data. The findings lead to an advocacy of quite simple variants of a hybrid formula suggested by Lent and Dorfman that use the same data as Laspeyres-type indexes but substantially reduce their bias.