Purpose -The purpose of this paper is threefold: first, to present a framework of five "facets," i.e., distinct but complementary ways in which the observed appeal of a consumer shopping site's features can potentially be generalized across product/service domains (the authors call this framework the feature appeal generalization perspective); second, to determine if and how observed feature preferences for consumer electronics, bookstores, and sites "in general" generalize across domains; third, to test hypotheses about the impact of frequency of domain usage upon feature generalizability. Design/methodology/approach -Via an online survey administered in a controlled laboratory setting, 313 respondents evaluated 26 website features in three domains (books, electronics, general) for a total of 24,414 preference judgments. Findings -Two facets, individual feature values and within domain evaluative dimensions, revealed minimal generalizability, while there was moderate comparability across all domains in between domain feature correspondence. Personal preference elevation could be generalized between books and general, but not between these two and electronics. Differentiating dimensions showed that preferences were not generalizable from electronics to books and general because consumers wanted electronics features to provide "flashy sizzle" and books/general features to give "comfortable safety." As hypothesized, patterns of generalizability coincided with frequency of domain usage. Research limitations/implications -Practitioners should not apply published studies of feature appeal to their domain of interest unless those studies directly analyzed that domain. Scientists should incorporate all five facets in modeling what attracts consumers to commercial websites. Originality/value -This is the first multidimensional analysis of the generalizability of site feature appeal across business-to-consumer product/service domains, and the first to propose this integrated evaluative framework with its unique facets.