Case Study
AI Recommendation System & Price Optimization
Client : HelloMolly
HelloMolly.com is an online retailer of womenswear. The company was established in 2012 and had quickly grown to become a leader in womenswear fashion. The company ships to more than 130 countries and have offices in US, Australia and China.
- 35% increase in customer engagement
- 25% increase in revenue
- Increase in sales resulting from personalised recommendation
Challenge
The company was looking to provide customers with a more personalized targeted experience and in turn significantly increase revenue and customer engagement from their marketing efforts. Online customers expect high level of personalisation and with their product offering so vast, delivering a personal and relevant shopping experience was very challenging. The online retailer wanted their years worth of data to be leveraged in order to improve the key area of their business.
What did Aegasis Labs do?
We built an AI powered Recommendation System and Price Optimization system that could offer highly personalised shopping experience to the customers.
We accessed and ingested the past historical data into our Artificial Intelligence System. Using this data we built a AI powered recommendation system which learned from past transactional and behaviour data of each customer. We also utilized our Customer Segmentation models to create data-driven customer segments for HelloMolly. Our bespoke models also included personalisation models such as in-market predictions, lifetime value, style preferences, pricing and churn.
Our AWS powered AI System, with bespoke algorithms and models, was integrated with their online platform to perform realtime product recommendations, optimization and offer a highly personalized shopping experience. A realtime dashboard was also provided which the retailer could access to track different KPIs, predict future trend changes and access other actionable intelligence and insights from different personalisation models.
Technical Bit
Our team specialises in Deep Reinforcement Learning which is a branch of Machine Learning used for optimization and control problems. Realtime dynamic recommender systems need to have two important charactersitics which other traditional Machine Learning and AI models cannot offer.
- Dynamic interactive nature between the users and the recommender systems
- Ability to take in long term effects and rewards and learn from them to improve performance
Static AI Recommendation systems need to be manually retrained when new data is collected. We designed a system which can learn as it interacts and gets feedback thus removing the need for manual retraining for new data. This dynamic nature of the system allows it to adjust according to seasonal changes, tastes, styles etc and constantly learn how the user’s shopping behaviour and taste are changing so that it can change and offer personalization accordingly. Our AI solution also offered recommendations in order to increase the long term customer engagement and conversions rather than just focusing on short term gains.
Results
With our AI powered solution, we helped the online retailer increase customer engagement by 35%. This helped increase the conversions by 25% thus increase overall revenue and profitability.
Our work with HelloMolly provides an insight into how Aegasis Labs helped them become an AI powered business and ultimately increase revenue and profitability. To learn more about how we can help you with increasing sales and company performance using predictive analytics and machine learning, have a chat to one of our directors. We’ll tell what’s possible, with no hard sell.
