The Challenge
An industrial technology manufacturer needed its website search to help visitors find products and technical information across web pages and PDF documents. Most searches were for content, making the ability to retrieve and rank technical material particularly important.
Search results frequently appeared in an inappropriate order or did not adequately match the query. Difficult cases included exact and partial part numbers, language variants, multi-word searches, and expressions involving units of measure.
Visitors came to the site with different goals. Some were evaluating products, while others needed specifications, manuals, or other technical support. Search needed to account for both the words entered and the type of information the visitor was trying to find.
These requirements called for careful treatment of different query types. A visitor entering a part number needed precise matching, while someone searching for a technical topic needed relevant content surfaced from a broader collection of documents.
Innovent’s Approach
Innovent reviewed the search implementation to understand how its configuration affected matching and result ordering.
The team examined the reported problem areas and tuned relevance to better accommodate the manufacturer’s content and search requirements. The work addressed the specialized language and identifier patterns that made these searches difficult, rather than treating every query as a simple keyword lookup.
Innovent also identified opportunities to make multi-word matching more precise. Broad matching could return content that addressed only part of a query and introduce less useful options into filtered navigation. The team recommended adjustments to matching requirements and further testing against representative searches.
Looking beyond the immediate tuning work, Innovent recommended ways to incorporate content classifications, query intent, and visitor interaction data into future relevance improvements.
Innovent also advised on a methodology for measuring relevance over time, giving the client a way to evaluate future changes against representative search needs.
The Result
The review and tuning significantly improved search relevance, addressing problems with matching and result ordering across technical content and product searches.
The engagement demonstrated the importance of aligning search behavior with the information people are trying to find, from a specific part number to a technical subject, and provided guidance for evaluating search quality as the implementation evolved.