For one leading retail group, success and double-digit online sales growth came with a cost: a search and recommendation experience that couldn’t keep up, threatening to send customers to competitors instead.
The effort vastly improved the company’s discovery and personalization functions; they knew that in order to succeed in this era of e-commerce their product search and recommendation systems needed to be much more intelligent than what they were. They turned to ALETERIS to rebuild their product discovery with AI.
Challenge
The biggest challenge was that Verano Retail Group was not utilizing AI properly in its online store. Too much of the product catalog relied on basic keyword search, which meant that customers often couldn’t find what they were actually looking for. Personalized recommendations were also not available, and merchandising decisions could only be made after sales data had already come in. This was holding Verano back; they knew they could capture more of the market if they had the ability to understand customer intent in real time. The work addressed three critical issues for Verano Retail Group:
Improve search relevance and product discoverability:
The teams focused their efforts on a few of the highest-value search signals in order to review the current search process, identify gaps in the catalog infrastructure and analytically understand query-to-purchase variability.
Determine the right recommendation strategy:
With thousands of SKUs across categories, Verano Retail Group needed a proper method to predict and surface relevant products. Using a mix of collaborative filtering and behavioral embeddings, the teams defined appropriate recommendation logic by mapping actual versus predicted engagement on the most important product segments.
Optimize the funnel for perfect product-to-customer matching:
The diagnostic determined the stressors that affected conversion and average order value. The teams focused on resolving issues related to higher-than-normal bounce rates and cart abandonment, which stressed the entire shopping experience and led to lost sales.
Solution
The solution ALETERIS came up with combined cutting-edge machine learning with real-world retail practicality. Everyone knew that the search and recommendation systems had to be rebuilt, the real challenge was doing it without disrupting the live storefront in a negative way. The solution was to introduce a semantic search engine and a real-time recommendation model, while providing merchandising teams with dashboards to tune results.
Results
All product, search and behavioral data in a single spot. No need for manually curating featured products or continually updating merchandising rules by hand.
Every customer gets a personalized shopping experience, not a one-size-fits-all catalog page. We understand that every shopper and every visit is faced with unique intent and context, and therefore requires specific results and direction.
By the numbers, the effort:
Increased conversion rate by 23%
Increased average order value by 18%
Reduced search abandonment by 50%
Improved recommendation click-through by 95%