SCENARIO: Fashion Retail Recommendation System via Multiple Clustering Approach

The competitiveness of retailers strongly depends on the conquered reputation, brand relevance and on the marketing activities they carry out. The latter aspect is exploited to increase the sales and thus a retailer, through marketing, should be able to encourage customers to buy more items or more valuable items. CRM solutions on the market have a very traditional approach based on popular items or bundled offers, similar items or featured items and therefore often leave out the important customer centricity in any marketing strategy.

Current solutions on marketing provide a too general approach, based on most popular or most purchased items, losing the focus on the customer centricity. Snap4City technology has been exploited on real retail company Tessilform, Patrizia Pepe brand in the context of the Feedback project founded by Regione Toscana, in which a recommendation system for fashion retail shops has been designed and developed, based on a multi clustering approach of items and users’ profiles in online and on physical stores. The proposed solution relies on association rules mining techniques, allowing to predict the purchase behavior of newly acquired customers, thus solving the cold start problems which is typical of current state of the art systems.

The solution has been validated on real retail company Tessilform, and it has been validated against real data from December 2019 to December 2020, showing that the use of the proposed recommendation tool generated stimulus to the customers which brought to an increase of buyers’ attention and purchase increase of 3.48%.

The solutions proposed has demonstrated to be functional also in the presence of low number of customers and items, and when suggestions are mediated by the assistants as happen in the fashion retail shops. Moreover, the proposed solution addresses and solved lacks and issues which are present in current state of the art tools, such as also the cold start problems in generating recommendations for newly acquired customers, since it relies on rules mining techniques, allowing to predict the purchase behavior of new users. Our solution is also GDPR compliant, addressing the current strict policies for users data privacy, solving one of the main issue for managing users’ demographic details.