We create a new database of retail prices in Argentina with over 10 million observations per day. Our main novel finding is that, different from Kaplan, Menzio, Rudanko, and Trachter (2016), chains, rather than stores, explain most of the price variation in our data. We show this in three ways: (a) Even though chains have on average 158 stores, there are on average less than 2.5 unique prices per product by chain; (b) Among products that change prices in one store, the probability that other stores of the same chain also change the price of the same product in the same day is 2.4 times the probability for other stores of any chain; and (c) A formal variance decomposition shows that only 28% of the price dispersion (for the same product, day, and city) is explained by stores setting different prices within a chain. This finding is relevant for retail-pricing theories since there are significantly fewer chains than stores, which matters for the degree of competition in the market. This paper also studies the heterogeneity in price changes and price dispersion across product categories.