Accenture employees turn to their internal search several times per second to find knowledge and experts to deliver better solutions for their clients. They also search for quick answers to internal personnel or support questions. But they often abandoned searches, frustrated with results.
Accenture’s Search Excellence team knew users were dissatisfied, but they lacked the visibility and tools to measure and influence search rankings. Content-side improvements were helping, but with dozens of sources and millions of items, the pace was too slow. Accenture’s internal IT organization decided a new approach was necessary to get a step-change.
Tapping into Accenture's rich content with a new search capability
Accenture managing directors Jason Warnke and Chris Crawford discuss how the new Accenture-developed enterprise search capability better serves our clients and our people.
A global Search Excellence team came together to rethink how to develop a better search solution. They knew core search engine technology was important, but installing a new product was not the solution. Rather, the solution needed to be a combination of measurement, analysis and technology, wrapped in ongoing relevancy improvement processes.
The Search Excellence team focused its initial efforts on defining analytics to use in continual search improvement processes. The team needed immediate and repeatable metrics of search quality, not only to measure user satisfaction, but to evaluate possible relevancy tuning changes.
The team created unique scoring tools that could quickly rescore historical searches and evaluate a curated set of “golden queries” to predict the impact of a proposed relevancy change. These scoring mechanisms were core to the iterative improvement process.
Unable to find an out-of-the-box enterprise search tool that would give Accenture the ability to understand and tune ranking, the team turned to building a solution using an open-source search engine and adding layers specific to Accenture’s business. Adding a new layer to the indexes allowed the team to improve, normalize and enrich the metadata across multiple data sources. The additional layer also enabled modifications to the tuning, without lengthy content re-crawls.
The resulting technology solution consisted of five main components: a crawler, enhancement layer, search engine, user presentation, and reporting and analytics.
Accenture’s new search capability geared to its needs is now live, providing more relevant searches and encouraging use of Accenture’s extensive knowledge resources to better serve clients. In addition, the capability is better, faster and more cost-effective in the following ways:
Better Search results are more relevant and accurate. “Abandoned searches,” a user leaving the search results page without clicking, decreased by 25 percent, relevancy scores for the benchmark search terms are steadily improving and anecdotal feedback has been overwhelming positive.
Faster – Users get results in under a second, delivered five times faster on average. The search team can test relevancy changes on-the-fly, rather than after a day-long crawl.
More cost-effective – The new solution runs with fewer and smaller servers. It uses open-source software that is proving simpler to upgrade and manage.
Accenture’s search capability breaks new ground by taking an analytic-driven, continuous improvement approach to search, supported by innovative use of open-source search technology coupled with custom extensions and tooling.