Evolutions of personalized recommendation services
Technological advances have made it possible to evolve e-commerce site recommendations into three main stages:
Each module is specialized in a type of recommendation service such as Cross-sell, Upsell or Best Sellers. You manually configure products to recommend by product or business rules. You must know the characteristics of each product and evaluate the products to associate in each banner. It is relatively effective for products you know well. But the efficiency is much more random for products that you know less well. When you edit an item in your catalogue, you must analyse all potential changes to be made manually. This method can become very tedious if you have a lot of products and a strong rotation. The relevance of the recommendations depends on your knowledge of the products. The acquisition cost of these modules is low because they are relatively simple to develop but the management cost can quickly become high when your catalogue increases. They are highly deployed by small e-commerce sites.
Introduction of Artificial Intelligence
Artificial Intelligence has made it possible to automate the processing of each recommendation. Some results are updated every day, or even at each visitor’s click. The recommendations become personal depending on the browsing behaviour of each visitor. Artificial intelligence algorithms based on the “wisdom of the crowd” come up with amazing recommendations that you never thought of. They are on average much more relevant than manual associations. Recommendation Service algorithms help you understand how your customers’ behaviours change and how each product contributes to your sales success. The first solutions were generally particularly complex to configure. They required to understand the options and parameters of each algorithm with installation manuals that could reach hundreds of pages. The implementation of business rules to highlight promotional or destocking products requires manual associations incompatible with predictive algorithms. The relevance of the recommendations depends on the configuration of the algorithms. The cost of acquiring the first solutions was high because the learning algorithms are complex, and the cost of integration and management can become very high. They are deployed by large e-commerce sites with a budget and qualified staff.
Simplification of service management
The most significant evolution for the merchants of Artificial Intelligence solutions is the simplification of the management of personalized recommendation services. Automation of treatments is no longer limited to recommendations. It extends to service management, simplifying the integration, configuration and updating of services.
MyDreamMatch is part of this new generation to reduce the cost of managing the service to the strict minimum. You do not have to worry about choosing the best algorithm for each service, and each page. MyDreamMatch does it for you, taking into account your industry. We have implemented a simple method of highlighting products, compatible with predictive algorithms. You place each recommendation banner on pages and at the location of your choice without any computer knowledge. We combine the seven essential recommendation services across all pages to support each customer throughout their entire buying journey. The relevance of the recommendations is high thanks to Artificial Intelligence. The acquisition cost is low because the market is important, and the cost of integration and management is minimal. The time has come for personalized recommendations for ALL!
Software engineer from Supélec Paris, I worked as a developer, architect and then manager for many years at the Research Center of IBM and then in the Cisco European organization.
Passionate about Artificial Intelligence, I decided in 2015 to create my company MyDreamMatch in order to make personalized recommendations accessible and easy to use by the largest number of web sites.