Russian Agricultural Bank (RusAg, the Bank) became the winner of the RB Digital Awards 2023 in the "Marketing and Sales" nomination for the use of new technologies in the field of marketing and sales. A high rating from experts was received for the recommender system developed for the farmer’s products marketplace ‘Svoe Rodnoe’.
‘Svoe Rodnoe’ is the largest online marketplace that unites more than 10,000 farmers across Russia and helps to find the nearest producers and order natural products from them directly and without overpayments.
The product range size and categories is constantly growing: over the past year, the number of products has increased 3 times - from 35 to 104 thousand items. This is main trade advantage for potential buyers, but at the same time creates natural barriers to finding the right products and increases decision time. It should be noted that the logic of displaying product catalogs on the ‘Svoe Rodnoe’ platform implies a geo-dependent issuance of farmers and their products. This means that a visitor from Moscow will see one set of products, while a user from Vladivostok will see another. At the same time, farmers can have an offline sales point in a particular city or deliver products through federal logistics companies throughout the country, which determines the geography of accessibility for end consumers.
The recommender system is based on 3 main approaches to the formation of the best offers:
1. Based on the ranking / ratings of goods;
2. Based on product categories (item-based);
3. Based on user preferences (user-based).
The uniqueness of the recommender system lies in the introduction of recommendation algorithms, taking into account the factor of geo-dependent product delivery at all stages of the path that the buyer goes through from the moment of viewing the catalog to placing an order. are taken into account preferences users Svoe Rodnoe platforms by regions, which allows the recommender system to draw the attention of customers to the most popular products in the regions. This, in turn, allows you to optimize the decision-making stages, offering the most relevant products and reducing the time to add an item to the cart. The implementation results have significantly improved key product metrics:
The uniqueness of the recommender system lies in the introduction of recommendation algorithms taking into account the factor of geo-dependent product issuance at all stages of the path that the buyer goes through, from the moment of viewing the catalog to placing an order. The preferences of users of the ‘Svoe Rodnoe’ platform in the context of regions are taken into account, which allows the recommender system to draw the attention of customers to the most popular products in the region. This, in turn, allows you to optimize the stages of decision-making by offering the most relevant products and reducing the time to add an item to the cart. The results of the implementation made it possible to significantly improve key product metrics:
the conversion of transitions to the catalog of a particular seller increased 4 times;
the conversion of adding goods to the cart increased by 18%;
returns increased by 12%.
In the near future, it is planned to optimize the accuracy of the models for various customer segments, which will allow the Bank to realize a greater potential for growth in indicators.