Every March, college basketball fans around their world lose their minds. Our team got the opportunity to work with Google and the NCAA to create a responsive website that used machine learning to create live March Madness predictions all tournament long.
View more New Products
In this case, we had three clients, Google, NCAA, and our agency partner, Eleven, who was focused on creating the campaign’s television commercials. The tournament was fast approaching and little attention had been paid to the online hub. A college hackathon was being planned and the outcome of that needed to influence our product and provide content. Additional challenges we faced were around the need for editors to upload content throughout the tournament while still abiding by Google’s rigorous security standards.
This project required a lot of fluidity as the general concept was established, but the look and feel as well as the content was still very much in flux. Our solution was to use Google App Engine with Angular 4 to create a site that tells the story of the campaign while also being a utility for those interested in the insights.
Many different types of predictions were made specific to metrics like “number of combined rebounds in a game” or “percentage of time leading”.
It was very important for our client to have exceptional performance in this campaign, so our engineers refined the page load processes until we were able to achieve a 98/100 on mobile and a 100/100 on desktop.
This project started with a discovery phase to better understand what type of Insights might be generated. However, we found that the answer to that question was “we’re not sure yet and need to be able to adapt”. That in mind, we designed a system of graphs that could easily swap out data points at a moment’s notice while still displaying quantitative measurements.
We designed eight different types of graphs for the editorial team to choose from when publishing Insights. You can check out the Hub here until 5/1/19.