Several of Rogers websites – macleans.ca, chatelaine.com, advisor.ca and canadianbusinesss.com – had suffered due to poor search functionality. Rogers had selected Endeca’s search and information access software to resolve the issue and needed a firm that could implement Endeca effectively. RealDecoy was recommended by Endeca and selected to implement the new search solution across the four sites.
THE CHALLENGE
Poor search results
The percentage of visitors who were leveraging the search functionality on the four web properties was low. Search results weren’t proving particularly relevant. Articles returned were often out of date or were not relevant to the search term.
Inability to access metadata
On the technical side, Rogers was having difficulty indexing the data on its site for the purpose of yielding worthwhile search results. Because its system simply used a Web crawler to search for data, there was no way for the crawler to access the metadata associated with a particular article and it was impossible for the search engine to rate the relevance of certain keywords.
If a visitor was searching a site for an author of a particular article, the site’s search engine had no way of knowing the difference between the author, say Marc Chrétien, and an article’s subject, for instance, the former Prime Minister Jean Chrétien. This contributed to the poor search results which were frustrating site visitors.
OUR SOLUTION
Filtering results through a content management system
RealDecoy was able to solve the poor search results problem by switching the search engine’s data acquisition from a Web crawler approach to a direct integration between Endeca and Rogers’ content management system (CMS).
Once we had access to raw unstructured data and elements such as taglines, keywords, authors and publication dates, we added a custom template to the CMS. This template allowed web editors to update the site simply by clicking the ‘publish’ button and also generated an Extensible Markup Language (XML) file that would be used by the search engine. This allowed the search engine to filter information based on metadata in the XML file.
A scalable template
RealDecoy developed a scalable template so that when Rogers is ready to convert other properties to Endeca as a search engine, it can use the base infrastructure now in place. This gives Rogers the flexibility to build out further tools related to Endeca in the future.
Our work on Rogers’ media properties left the client so pleased that we were brought in to implement the Content Acquisition and Syndication Hub (CASH).