x

Search the Findwise web

Findwise + EVRY. Creating a powerhouse in Digital, Advanced Analytics and Cognitive.

Enterprise Search at a global technology company

Personalized search results in Netflix-style user interface

Challenge

One of the world’s largest technology companies, with headquarters in Sweden and over 100.000 employees globally, had an enterprise search solution for over a decade. The aim of the search capability was to improve information re-use among the workforce and thus support the company’s knowledge management efforts. Perceived relevance of search results among the end-users was however generally regarded as unacceptably low.

Users were drowning in too many hits of various importance to them. The previous solution was based on proprietary search technology which limited the understanding of the root causes of the issues.

One of the general problems with “black-box" search software is that it rarely works well out-of-the-box mainly due to the heterogeneity and complexity of the information sources made findable. In this case, the software was not configurable enough to make it possible to provide the individual end-user with the kind of contextualized search experience that would have been needed to improve perceived relevance. The complexity of the information landscape at the company and the level of manual maintenance required to keep the solution up and running resulted in a solution, which did not meet expectations from a business value, cost or technology perspective.

Solution

Together with a team from the customer, Findwise designed, implemented and rolled out a new enterprise search solution based on Findwise I3 and Elasticsearch, the open source search and big data platform.
The I3+Elasticsearch-based solution opened the door to using enterprise search not only as a centralized search portal accessible from the intranet, but also as a platform which allows crowdsourcing by internal users allowing the solution to be collaboratively improved.

Third-party application owners at the company can connect to the platform to use information from other applications swiftly for their own purposes without compromising information security. In addition, information owners can make their own information findable via the solution with little help required from the search team. This opens for a whole new set of search-enhanced applications to be developed for specific use cases with much shorter time-to-market than before. The need for use case-specific search engines is lowered along with the risk of costly shadow IT initiatives in the search area.

Result

Included in the solution is a Netflix-style user interface (in addition to traditional Google-like list) to visualize the vastness of available corporate information and invite the end-user to drill down to various types of information like “videos”, “courses”, “people”, "buildings", "knowledge objects", "directives", "conference rooms", “product manuals” and “instructions”.

With built-in support for several relevance models, one for each use-case, the models are adaptable, i.e. can be changed on the fly by an application owner. From start, the end-user’s geographic location and organizational belonging is automatically considered for boosting information of value for the end-user’s context. Personalization is built in to I3 and can be adapted with more contextual signals if deemed useful and when available from the underlying sources.

User satisfaction levels for the global enterprise search portal is rising with first evaluation going from 59% (2015) to 72% (2017) satisfied end-users.

“Earlier when we wanted to promote our product in the enterprise search portal, we were forced to ask the search team to add a recommended link as the product information was never showing up properly among the organic hits. With the new tool, we can actually trust that the information we have will be promoted when people look for it. That was definitely not the case earlier.” – Product manager at the company.

Netflix-inspired user interface.
User interface that is promoting filtering in various ways. Selected filters are visible in the search box.
More information about a particular document, page or object is provided in a “Show more”- view.