‍BrandAlley is a leading UK e-commerce platform for designer and luxury fashion with 5M+ users that adds over 30 thousand new products per month, and runs 25 new flash sales every week. To recommend products to users, BrandAlley was spending a lot of overhead manually curating product blocks, and their recommendations still fell short of the kind of nuanced, up-to-date personalization they needed to remain competitive.

BrandAlley were looking for an automated system solution that would:

• handle the cold start problem - creating real-time, personalized recommendations for newly added prod
• generate real-time recommendations for all products, and
• scale to handle BrandAlley’s sheer volume of inventory and customers.

In 2024, BrandAlley started working with Superlinked, and everything changed.

With Superlinked's vector-search-powered recommendation framework, BrandAlley was able to transform their entire shopping experience, achieving a 77% conversion rate increase, and a 68% higher average order value. Let’s see how, below.

‍

BrandAlley’s challenge

BrandAlley was facing a critical juncture that threatened their competitive position in the UK fashion market. Successful e-commerce was increasingly personalized, and good personalization required robust technical solutions.

Market Reality - Personalization is King

BrandAlley operates in a fast-paced, high-volume retail environment, where the vast majority of consumers (76%) prefer buying from companies offering personalized shopping experiences; personalization drives 78% higher repeat purchase and referral rates (McKinsey, 2023).

BrandAlley’s personalization needed improving, and fast.

Operational Hurdles

• BrandAlley’s one-size-fits-all sales block approach was costing them time and lost profits
• countless hours manually curating 100+ homepage sales blocks for a dynamic catalog of 32k new products monthly and 25 weekly sales
• missed opportunities for real-time personalized recommendations and upsells
• new products, because they had no user interactions to base recommendations on, suffered from poor visibility, took too long to surface to potentially interested customers

Technical Complexity

Overcoming these operational hurdles demanded more than a basic recommendation algorithm or merely feeding all their unstructured data into an LLM. BrandAlley needed: 

• sophisticated embedding techniques to deal with complex real-world, multimodal data - product images, text descriptions, prices, and inventory status
• real-time processing of user behavior, and sub-second response times
• to scale their recommendation system to handle millions of users and a constantly changing inventory

‍

The Superlinked Solution

BrandAlley turned to Superlinked to help them set up a comprehensive, real-time, personalised recommendation system that meets their needs. Superlinked’s team designed a solution that relies on real-time streaming of user interactions with the BrandAlley UI via Streamkap, and uses Redis as a robust vector search solution for storage and real-time retrieval. 

‍

The solution:

• provides seamless multi-attribute, multi-modal data integration. By integrating data schema definition and the ability to create matching embedding models, Superlinked’s framework lets you combine text, image, and categorical data with adjustable weights, to generate accurate product recommendations.‍
• continuously incorporates and processes real-time user behavior data, including views, purchases, and searches, to update the user index, ensuring your recommendations reflect current user interests and preferences.‍
• creates personalized (customized) recommendation endpoints you can quickly deploy to A/B test, and fine-tune through weight adjustments, allowing teams to optimize for specific business goals while maintaining performance.

‍

With Superlinked’s platform, BrandAlley was able to implement:

• Reliable, highly relevant, AI-driven recommendations, instead of manually curated product blocks
• ‍Real-time updates to make sure recommendations reflect current inventory and user behavior - the Superlinked system indexes all product updates and user events within seconds‍
• A New, Personalized "For You" Tab - with real-time updates based on user behavior and preferences‍
• Smart Product Pages, with intelligent "You might also like" sections that adapt in real time to user behavior - producing significant performance improvements in A/B tests‍
• Dynamic Cart Upsells - real-time cart suggestions with sub-second response times - leading to increased upsell conversion rates

‍

Results

The implementation delivered substantial improvements across all key metrics.

Operational Improvements:

• Reduced manual curation time by 90%
• Achieved sub-second recommendation response times
• Increased new product discovery by 125%

Business Impact:

• 77% higher conversion rate
• 68% increase in average order value
• 60% growth in revenue per user
• 43% more items per order

‍

The Superlinked Advantage

• Solves the multimodal embedding challenge: BrandAlley can now effortlessly and natively combine complex data types – product images, text descriptions, user behavior, and inventory data – into a unified view that powers more accurate and relevant recommendations.

• Provides real-time personalized customer experience: By delivering highly personalized product discovery, BrandAlley customers find relevant products faster, leading to increased satisfaction and deeper platform engagement.

• Grows business: Better product discovery mechanisms and improved customer retention have directly and dramatically improved BrandAlley’s profits, and created sustainable growth patterns.

• Future-ready, scalable performance: BrandAlley now operates on a future-ready technology stack that scales effortlessly with their growing user base and product catalog. Superlinked’s real-time processing capabilities ensure consistent performance even during peak periods, while its flexible architecture permits continuous innovation and feature expansion.

In short, Superlinked’s technical foundation positions BrandAlley for long-term success in a highly competitive and increasingly personalized e-commerce landscape.