TURF Analysis

Because the most popular items often appeal to the same people, so adding the second- and third-most-popular item brings in few new customers. TURF maximizes unduplicated reach: an item is valuable if it covers people no other selected item already covers. The classic illustration is flavor selection, where the top individual choices may all be variations beloved by one segment, leaving other segments unserved. By optimizing the set rather than ranking items alone, TURF tends to recommend complementary variety that broadens coverage, which is usually the strategic goal of a product line.

Reach is the share of respondents who find at least one item in the chosen set appealing — an unduplicated count of customers covered. Frequency is the average number of appealing items per respondent, which indicates how deeply or redundantly the set engages those it reaches. A set can have high reach but low frequency (broad but shallow) or lower reach with high frequency (a loyal core that loves many items). Most TURF decisions optimize reach first, then examine frequency to judge redundancy, engagement, and the risk of cannibalization within the assortment.

Naive TURF evaluates every subset, which explodes combinatorially and becomes infeasible beyond a couple of dozen items. Serra showed that the selection is a maximum-coverage problem expressible as a binary linear program, so modern solvers return provably optimal assortments even for large candidate pools. When problems are extremely large or when building full reach curves, a greedy heuristic that adds the item with the biggest marginal reach is a fast alternative with a known approximation guarantee for coverage objectives. In practice analysts use exact solvers when feasible and greedy or branch-and-bound otherwise.