This article presents a comprehensive examination of panel data for 1,847 consumers and 2,209 brands of "biscuits" (a total of 76,682 records) in which matching analysis is employed to define brand substitutability and potential product clusters within the overall category. The results indicate that, while brands performed as expected as perfect substitutes for one another, five subcategories of biscuits into which the brands were divided (chocolate biscuit countlines, chocolate-coated biscuits, filled biscuits, plain sweet biscuits, and savory biscuits) generally performed as separate products. Matching provided a graded measure of substitutability/nonsubstitutability of brands and products, and thereby contributed to their definition. (Contains 2 tables.)