Bringing Personalized Digital Experiences with Microsoft Azure Personalizer for an E-Commerce Company
Microsoft Azure Personalizer
Average Increase in Sales
Higher View to Submissions Rate
E-commerce stakeholders need to deliver engaging and relevant content to their users. E.g. product/services offer, media or ads. A highly competitive e-commerce segment needs to provide personalized and guided digital experiences to its users to get more leads and conversions. Irrelevant content is a highly frustrating experience to online users resulting in higher abandonment and page bounce rates.
E-commerce companies struggle with content highlighting, filtering, and placement for their web portal and mobile app. Ad placement and UI usability improvement is another area of concern for the e-commerce domain.
Our client was facing these issues which resulted in visitors dropping off, affecting conversion rates and sales. Their current marketing strategy based on A/B testing failed to deliver the desired insights to ensure higher conversion rates.
- Failure of A/B tests
- Personalizing content is a complex and dynamic problem
- Lack of robust cloud platform
- Siloed data
- Inability to derive insights from customer data
Smarter and Personalized Customer Experience With Reinforcement Learning Solution
Modern technologies elevate digital experiences with higher personalization and accuracy. Machine learning-based content delivery and placement help to boost user engagement and user experience on websites and apps. Integration of RL models for content prioritization brings precise results as they mature with no prior dependency on historically labelled data and deliver accurate predictions over time.
With a clear understanding of the requirements, our experts designed ML solutions to help the client. Upon analyzing the client’s technology stack, we suggested a reinforcement Learning-based recommender system using Azure Personalizer. It helps to determine the best content to serve based on content relevance and reward score.
The reward score was automatically computed between the -1 to 1 range, where -1 indicates low relevance of content and 1 denotes high relevance of content to user intent. Our RL Personalisation model consumes the feedback (reward/punishment) based on the actions that the user takes during a session, which eventually updates the model weights.
With all the objectives in place, we determined a strategy with periodic KPIs to ensure seamless execution of the project.
Our Microsoft Azure experts:
- Designed a RL Personalization model to rank and prioritize content
- Enhanced model performance with automatic reward submission based on content relevance
Our reinforcement learning model considers the following inputs to rank content:
- User information and features
- User actions and features
- Reward score based on each user action
Attri’s customized Azure Personalizer solution addressed content highlighting, content ranking, personalized offers, and UI usability improvement for our e-commerce domain client. The solution helped our client to deliver tailored content with the appropriate depth and tone by differentiating the audience.
The model attains greater accuracy with reinforcement learning-based reward scores than a vanilla recommender system would have given. This mechanism also ensures a fair degree of explicability for suggested ranks. The solution is scalable to serve other extended use cases such as content filtering, live events, and ad placement.
While delivering our ML solutions, we ensure that the client's in-house team gets complete knowledge and support to execute the project flawlessly. Our experts provide UX training, knowledge transfer, and comprehensive documentation to deliver the best experience with the solution.
- Enhanced customer experience with personalized content.
- Improved brand engagement through preferred channels
- Better results with more leads, conversions, and reduced bounce rate
- Improved brand affinity
An e-commerce company leveraged Attri’s expertise to delight their customers with personalized experiences using Microsoft Azure. Attri implemented an ML solution to determine content ranks and placement on the clients’ website and app. The solution handles content ranking, filtering, and prioritizing for a more personalized experience.