Approach
Search audit & recommendations
In collaboration with a design agency, Dovecot audited current search performance by analyzing years of search logs and site analytics . We also created a set of most common search scenarios for key personas to design a new relevance model for the search engine. By developing “search patterns” – common search categories and the typical results expected for both internal and external users – we pinpointed problems within the current search experience. We evaluated search relevance against these patterns to identify opportunities for improvements to the relevance and ranking model.
Taxonomy integration
In a prior engagement, Dovecot revamped the institution’s enterprise taxonomy. For this project, Dovecot worked with stakeholders to implement the revamped taxonomy into the web content management system so that content could be tagged with subjects and other key metadata. This would allow metadata to flow from internal creation to external publishing, reducing effort for web managers. It would also provide a rich set of synonyms to be incorporated into the search to improve results.
Search UX and faceted filtering
We worked with the design agency to craft the new search user experience, including improving how search results are displayed and adding faceted filters. After implementation, the newly integrated taxonomy was turned into filters so users can sort results by content type, publication, subject, country, and other key data points.
Auto-classification implementation
To make the new relevancy model and filters work, accurate and consistent taxonomy tagging was critical. To help authors apply rich metadata, we worked with the team to integrate the enterprise auto-classification solution into the WCMS. This included defining classification rules for various web content types and publications.