Recent empirical evidence shows the Law of One Price does not exist in online environment despite the unprecedented amount of price transparency. The failure of Law of One Price has been attributed to search costs, brand loyalty and other kinds of “market frictions” that enable retailers to keep some of their customers even if they don’t charge the lowest price in the market. While the presence of such frictions has been carefully studied, little is known empirically on how they affect consumer demand structure that is at the core of retailers’ pricing and marketing strategies. In this paper, we identify the impact of market friction on demand structure by empirically demonstrating a well-known insight from economic theory that market frictions could lead to a kink in consumer demand function. Using a unique dataset on prices and demand collected from Amazon and BarnesandNoble.com (BN), we first demonstrate the existence of consumer segments that demonstrate significant amount of frictions against shopping for lowest prices. Then we incorporate this finding and build an empirical model to show that significant jumps in price elasticity exist at points where price changes occur, which in turn, manifests as kinks in the aggregate demand function. Moreover, the jumps have opposite directions on Amazon and BN, indicating that market frictions have different implications across the two major online retailers. By examining kinks in demand functions, we contribute to prior empirical literature that has typically considered a constant level of price elasticity. We find that, on Amazon, price elasticity increases after a price reduction, suggesting that customers face low search costs for price information on Amazon or low brand loyalty toward BN. On the other hand, price elasticity decreases after a price reduction on BN, suggesting customers face high search costs for price information on BN or high brand loyalty towards Amazon. Further, an analysis of the differences in jumps in price elasticity for popular books compared to rare or unpopular books reveals that market friction is much higher for unpopular books. These findings suggest that online retailers have the potential to extract more value from the emergence of the Long Tail phenomenon.