What statistics could an online book store use? This is a theoretical question. 
If I had an online bookstore what kind of statistics would I keep.
The number of times a book was viewed is one example. Another example may be the number of visitors by country. 
What other variables are there which in different combinations with each other can be used to build charts, calculate curves and determine maxima(s) and/or minima(s) that will help to maximize profits.
The main objects in the database are books, users, transactions etc. 
For example: making a probability distribution chart based on past transactions of a person buying a book tagged with a specific genre, or making assumptions for investing in the purchase of new books based on the sale of previous years. 
 A: To learn about this you should study the website of amazon.com: http://www.amazon.com/      They obviously are using a lot of statistics!  In our time, storage is cheap (almost free), so you should store just about everything,  what matters is with what structure you store it.
Let us have a look at that website and see what statistics they obviously used.
Let us search for one book, I type in the search field: "Watson mathematical analysis"  (in parenthesis, a very good book which is still sold and read, dispite being more than one hundred years since first published!). The first hit is (page after choosing first hit):
http://www.amazon.com/Course-Analysis-Cambridge-Mathematical-Library/dp/0521588073/ref=sr_1_1?s=books&ie=UTF8&qid=1423485039&sr=1-1&keywords=watson+mathematical+analysis
We find: "Customers Who Bought This Item Also Bought".  To have access to that statistics, you must store all your customers and what they bought. 
You also find: "What Other Items Do Customers Buy After Viewing This Item?". So you must store the view sequence of all customers, and what they eventually bought on that visit.  This is information complementary to what you found on the first list. 
Below that you find:  "Your Recently Viewed Items and Featured Recommendations ".  For this item, storage is not enough, amazon in some way have found "recommendations", books they think might interest you (I find that very usefull, sometimes gems can be found in that list!)  How can amazon calculate recommendations?  Here I think, methods of multivariate analysis comes in. You can make a big matrix, customers as rows, books as columns.  This matrix is way to big to store in memory, can be compared to the "google matrix". But, as is the case with the google matrix, most entries will be zero (representing customers/books that didn't "meet" each other), so the matrix can be stored as a sparse matrix. Now all kinds of multivariate statistics methods can be applied, but the usual textbook representations of such methods concentrate on small matrices,  and our emphasis in using those methods will be very different. We can use
--- correspondence analysis
--- clustering (of clients, of books)
--- discriminant analysis
--- logistic regression (what is the probability of buying a specific book, for a specific client, given his view/buy history)
You can certainly find use for other methods.
Earlier times, amazon used to have a list of words important in the book, words that in some sense discriminates the book. This could be defined as words commonly used in the book, much more than their frequency in the literature as a total. I cannot seem to find that list now?  That used to be useful, as it could say quite a lot about the book!
Below that on the page, there is: "View or edit your browsing history". This can be very useful, if you want better automated recommendations, you can   delete books you arent interested in any more, or mark some books as especially interesting.  That way amazon can come up with better recommendations for you. 
So what should you do? continue this study of amazon.com website with statistical eyes!
