Only if your measurement is the number of products. If you were to represent each product as a point in a multidimensional space of features, and measure according to how well they were clustered, then the existence of Reader would improve the clustering of Google's products, helping to counter the existence of products like self-driving cars.
