Dynamic Yield’s Deep Learning Product Recommendations Generate More Revenue Returns


Dynamic Yield, the Experience Optimisation platform, announced the gradual release of its state-of-the-art, self-training Deep Learning Recommendations Algorithm, enabling brands to predict the next series of products a consumer is most likely to buy.

Today, product recommendations are an essential requirement for any eCommerce business looking to increase engagement, purchases, and loyalty. However, a consistent challenge for marketers and merchandisers has been determining which products among a massive catalogue of items to serve customers with various preferences and levels of intent.

Dynamic Yield’s Deep Learning-Based Recommendations instantly identify intent, even from the first session, to automatically match customers with the products they are most interested in or likely to buy, adapting as new data is ingested. The model employs the item2vec method, derived directly from its Natural Language Processing (NLP) counterpart, word2vec, to learn the products in a user’s browsing history, in-session activity, and trends seen across the site to recommend products each individual is predicted to engage with as they shop.

“Consumers have come to expect a high level of personalisation in online retail interactions,” said Liad Agmon, CEO of Dynamic Yield. “Our Deep Learning model exploits cutting-edge neural network technologies of natural language processing that are found to be extremely effective within the recommendations domain, providing a superior approach for predicting customer wants and needs.” 

Leading brands such as URBN Brands, OFFICE, GlassesUSA.com, and more are currently using Dynamic Yield’s deep learning algorithm to maximise the performance of their product recommendations. The advanced machine learning-powered strategy has already generated substantial double-digit uplifts in purchases and incremental revenue for Dynamic Yield clients.

Also Read: Make Way for A Podcast Hustle

Key benefits of the Deep Learning Recommendation model include but are not limited to:

Results optimised per user – The deep learning algorithm automatically determines the right subset of parameters for each user based on their behaviour, their position along the customer journey, as well as any trends identified across the site, eliminating the need to manually apply custom filter rules.

Rapidly trained and adaptive – The algorithm is constantly and rapidly self-learning based on a huge amount of behavioural and product data to power recommendation results that instantly identify customer intent, even from the first session, and are continuously refined as new information comes in.

Available within key digital channels – Apply the same advanced deep learning technology for delivering product recommendations predicted to drive engagement on the site within the mobile app as well as via email campaigns, which are tailored at the time of email open.