Uncover the latest AI trends in Appen's 2024 State of AI Report.
Resources
Case Studies

Search Functionality Improved With Search Evaluation From Social Network

The social network provider was able to improve its search capability in a cost-effective and efficient manner by partnering with us
March 27, 2020
Share

The Company

A major social network provider looking to improve the search functionality on its site—allowing users to successfully search for people, posts, news, and other items of interest.

 

The Challenge

A major social network provider needed to improve the search functionality on its site—allowing users to successfully search for people, posts, news, and other items of interest. The company was working with another vendor at the time, but that vendor was unable to meet the quality management and ramping demands of the project. Based on the success of a project within another group, this team decided to move forward with us to make the necessary improvements.

 

The Solution

The initial approach involved a pilot with 80 US-based raters measuring the site’s news search functionality. The client was immediately impressed with the our project managers’ expertise in search relevance evaluation, as they were able to grasp the task quickly and provide valuable feedback on the guidelines. In a matter of weeks, our raters were ramped up, exceeding the client’s quality expectations.

With the success of this pilot, the client wasted no time in engaging us for additional search relevance work, with projects focusing on social search, video and image search, trending topics, query classification, and other aspects of search. Some of these projects were moved to us from the previous vendor as our team was able to deliver results with higher quality, and others were brand-new projects as the client continued to experiment and improve its site’s search engine. Within one year of the initial pilot this program grew from 80 US-based raters on one search project to nearly 1200 raters from four markets working on fifteen search projects.

 

The Result

As a result of this project, the social network provider was able to improve its search capability in a cost-effective and efficient manner. Rather than spending time refining its own rating guidelines, the provider took advantage of our years of domain expertise and best practices, which also yielded higher quality data. The client now has a proven model for penetrating new markets and can do so in a fraction of the time.

Our years of experience in web search evaluation translate well to other domains, including social network search. The expertise demonstrated in this project includes continuous improvement of task guidelines, and the identification and management of strong raters. This all results in high-quality data that the provider uses to improve many aspects of its social networking site’s search functionality.

This client has also benefited from our ability to ramp quickly in markets around the world and meet the ever-growing demands of its search evaluation program. Our work from home model provides its clients to a large number of raters globally, without the overhead of on-site locations. Even within a single market such as the U.S., the client appreciates the diversity of rating feedback that the our model provides, as geographically-dispersed raters more closely reflect its user base than any single center of operations could.