The Challenge
A major retailer needed to upgrade its retail search and navigation platform while preparing for substantial catalog growth. The existing searchable catalog contained approximately 150,000 products, and the planned expansion would bring together internal merchandise and external vendor offerings to support as many as two million products.
The retailer also wanted product ranking to reflect business metrics such as conversion rates, revenue, and sales rank. These signals needed to support both general search results and merchandising placements for best-selling, recommended, featured, and related products.
Existing customizations complicated the upgrade. The retailer needed to replace custom code with standard platform capabilities while preserving the application interfaces and result structures used by its website. This required specialist search expertise across data integration, ranking, performance, and application compatibility.
Innovent’s Approach
Innovent worked with the retailer’s ecommerce team to plan, implement, and test the upgrade. The team assessed existing customizations, mapped them to standard capabilities, and tested the replacements to confirm that required functionality was preserved.
The work incorporated daily feeds of site statistics into the search index, making behavioral and commercial metrics available for product ranking. Ranking formulas could combine multiple signals rather than relying on a single measure, with separate business rules for different merchandising placements.
This approach connected search relevance optimization with the retailer’s merchandising requirements, allowing product ordering to account for both shopper queries and business priorities.
Innovent coordinated with the retailer’s developers to preserve the result structures expected by the website while introducing the new ranking capabilities. This reduced the application changes needed to adopt the upgraded search functionality.
The project also addressed vendor product feeds and catalog integration. Working with product vendors, the team established feed formats and incorporated their data so vendor products could participate in ranking and sorting alongside the broader catalog.
Incremental capacity testing evaluated the search architecture as catalog volume increased. Testing demonstrated support for more than three million products on a single search instance without performance degradation, providing headroom beyond the planned two-million-product catalog.
The Result
Query throughput per search instance increased nearly 150%. The deployment also moved from nearly 60 servers to 28, reducing infrastructure requirements while increasing search capacity.
The upgraded implementation brought metrics-based ranking and expanded merchandising capabilities into the retailer’s search experience. Catalog integration enabled internal and vendor products to participate in the same ranking and sorting processes.
The project was delivered ahead of schedule. Innovent’s work combined application compatibility, catalog growth, merchandising requirements, and performance validation in a coordinated retail search upgrade.