Recommender Systems have obvious influence in environments where data size exceeds the capabilities of any user to fully explore the available choices in the store (physical or on-line). Many algorithms and techniques have been used to help recommending useful and interesting items to users. If the user is unidentified, the process is even harder as there are no historical or other data to use as input. Association rules is a popular technique used for many purposes in Recommender Systems such as for building more robust systems, improving quality of recommendations; and even addressing fundamental limitations of recommender systems and, generally, large datasets, e.g. sparsity and cold start. At the same time, efforts have been made to fully understand if and how differently customers are behaving in an online and in a physical environment.This work tries to combine the two efforts. We use association rules to provide recommendations to customers, as well as understand who the customer is, what her needs are and what is her mentality when entering a physical store or the corresponding e-shop. To fulfill our goal, we used descriptive statistics along with Association Rules analysis of the POS transactional data on basket data level and historical data level.