Hi Mostater, Again, thank you for your contribution. At the end, I did something else. For your reference, here is what I did: First I summed up sales for the category of interest (beers). Thus, I ended up with a table having Beer Sales by Store By Month. Next, I calculated the euclidean distance between each store and all the rest. So I took the difference in Beer Sales for each month store by store against the rest, and squared that. After, I summed up the squared differences by month resulting in one number per each combination of stores. Finally I calculated the squared root of that sum, and that's the euclidean distance. Then, I just sorted the table by that distance and took the store pair (test-control) where that distance was the minimum for the test store. With this method, I ended up with 2 samples, one test and one control, with similar sales trend month by month over 20 months. Nice, isn't it?
... View more