If you are a content service provider like, or maybe smaller than, Yahoo! reaching out to vast audiences and offering them personalized digital content across devices and around the globe, then you need to read this to understand
In 2009, when Yahoo! signed a search deal with Microsoft, it was mutually agreed upon that Yahoo Search would be powered by Microsoft’s Bing. The data management teams at Yahoo! expected the larger datasets to run into petabytes. Already, the company had been collecting data about millions of users that visited its Web-portal through properties such as Search, Directory, Email, etc.
Rohit Chatter, Data and BI Architect at Yahoo! says, "Following the tie-up, the sizes of the datasets began to grow. There was a deluge of unstructured data to manage and analyze. It was mandatory that we implement a rugged and flexible BI tool."
Hitherto, Yahoo! employed a combination of Microsoft Analysis Services and Oracle’s tool for BI and Microstrategy dashboard for reporting. But the growth in digital properties (channels) and consequent data volumes, the high licensing costs of the proprietary tools was vexing.
The Development Phase
Yahoo! then put out 15 senior data architects backed by their teams to evaluate the user needs. The widening business scope had users from various departments (marketing, sales, research, executive management, etc) asking for features missing thus far. While business users wanted slicing and dicing of information, analysts wanted cube-based processing for complex report generation. The existing dashboard lacked functionalities such as downloading data for offline processing and had limited reporting metrics. Users wanted fast and flexible dashboards with metrics such as advertisements, click through rates, searches, publisher quality score, and visitors, to track performance of each channel and to understand their audiences better.
Since Yahoo! had played a major role in the Apache Hadoop project, it decided to develop the BI tool in-house using Hadoop resources and technologies such as HTML5 and CSS3. However, it retained investments in Microsoft and Microstrategy.
Some features the new tool incorporated were ad hoc querying, interoperability with existing applications, ability to generate custom-reports, flexibility in terms of slicing and dicing, scalability in terms of number of users, and the ability to handle huge datasets.
More on Yahoo!
Yahoo! created a private cloud to allow users to generate reports from anywhere, anytime. Following this, it carried out a load- and performance-testing of the tool just before rollout. To nip any quality issues in the bud, every business unit at Yahoo! had to follow predefined standards. A team of 20 and an additional three administrators are tasked with maintaining data quality.
Yahoo completed BI implementation in March 2011.
Business Yields of BI
With BI implementation, Yahoo! has acquired faster turnaround time to any business questions.
- Business queries that took 1-2 weeks to resolve now take less than two days.
- Yahoo! can generate reports based on three different cuts – the audience, the advertisers and publishers. Audience reports offer insights on the kind of traffic generated, advertisers reports help gauge the right pages to be advertised on, and publisher reports help understand which publisher is doing well in which area. Such granularity of reports enables analysis based on revenue, competition, and market trends.
- The tool facilitates competitive analysis. For instance, advertisers can identify popular keywords and accordingly reach out to audiences.
- Market trends offer insights on audience demographics helping Yahoo! to allocate advertising space appropriately.
- BI has increased the accuracy of reports and users can now auto schedule them.
- Improved revenue-reporting (based on number of clicks and amount of interaction that happens on each site) has helped Yahoo! to determine the accurate Click through rates (CTR) and RoI.
- BI implementation has enabled Yahoo! to focus on activities that boost revenue. “For instance, BI allows account managers and sales executives to analyze the huge data and cross-sell, up-sell, optimize campaigns and look at missed opportunities,” explains Chatter. “Through BI implementation, we have achieved 2% to 10% growth in revenue through advertisers, audience, and publishers,” he adds.
The Road Map
More BI case studies