The Challenge
An automotive parts retailer needed to modernize automotive parts search on its ecommerce website. The existing implementation presented problems with shopper experience, performance, and maintainability.
Parts discovery required search to account for product information alongside vehicle fitment, availability, and pricing. The retailer needed more relevant results, better autocomplete and navigation, and more flexible catalog updates.
The existing data model represented vehicle fitments as child records beneath products. This structure made queries and maintenance more complex. Faceting presented another challenge: vehicle fitment fields could contain enormous numbers of distinct values, even when a shopper’s selected vehicle narrowed the relevant choices to a small subset.
As search expanded into an application used in physical stores, new features and a doubling of indexed data introduced further performance and stability challenges. Subsequent traffic growth increased capacity requirements and made the cost of adding cloud infrastructure a concern.
Innovent’s Approach
Innovent redesigned the search data model, replacing the parent-child structure with a flat product model that represented vehicle fitment in carefully designed fields. This changed how fitment relationships were expressed and queried, reducing dependence on complex joins between product and vehicle records.
The team also reengineered full and incremental indexing from the retailer’s product database. The work addressed how catalog data was collected, transformed, and updated in search, supporting both complete rebuilds and targeted updates.
Relevance engineering addressed how product fields contributed to matching and ranking. Innovent worked with client staff to review representative searches and expected results, balancing exact matches with broader retrieval and using weighted ranking signals to reflect business priorities.
For fitment navigation, Innovent tailored facet processing to the structure and scale of the data. Vehicle-specific prefixes narrowed the facet values under consideration, while processing methods were selected to handle fields with very large numbers of values efficiently. This focused facet calculations on the comparatively small set of choices relevant to the selected vehicle.
The implementation included search architecture design for high availability, configuration across development, testing, and production environments, and support for storefront integration. Rollout and rollback planning, monitoring guidance, documentation, and knowledge transfer supported the transition.
Follow-on engagements addressed query and indexing performance, additional relevance engineering, and fitment facet processing. As the store application and indexed data grew, Innovent provided performance and stability troubleshooting. A later architecture and sizing review considered data volume, query growth, and configuration choices to help the retailer evaluate capacity needs and reduce cloud infrastructure costs.
The Result
The retailer gained a redesigned search foundation that represented product and vehicle fitment information without the previous parent-child query structure. Reengineered indexing processes supported catalog maintenance, while targeted facet configuration addressed the demands of unusually large fitment fields.
The engagement continued beyond the initial implementation to support search across ecommerce and physical-store applications as features, data volumes, and traffic expanded.
Innovent brought data modeling, relevance engineering, facet processing, and infrastructure planning together throughout that evolution. The work addressed the retailer’s immediate search requirements while supporting decisions about performance, scalability, and operating costs.