Sporting goods company and organizer of the largest road running circuit in Latin America, Track&Field® wanted to improve the experience of its athletes from the sign up to the conclusion of its events. That scenario required the company to undergo a process of digital transformation. Since 2016, Mesa has been working closely with Track&Field® in the development and evolution of TF Sports.
Our answer was developing a brand new platform, TF SPORTS, and its main challenge was supporting the user journey, right from the sign up to e-commerce and in-store purchases, delivery of race day running kits, to event attendance and the sharing of photos and results.
In addition to tracking the user, it was also important for Track&Field to develop algorithms that would inform the athletes' performance numbers in real time and process a large volume of event photos, identifying the exercisers in each one of them.
Through a long consultive process of racing analysis, user interviews and evaluation of the company's indicators, we were able to meet Track&Field's needs. For end users, we developed iOS and Android native applications, in addition to a responsive website where it was possible to access information about previous and upcoming events. Users could also register themselves and others, pay with bank slip or credit card, oversee the results of events and have access to personalized offers.
To keep track of the users' journey, a CRM tool was integrated with Track&Field's e-commerce, TF SPORTS and the POS of physical stores. The data collected by the CRM is used to support marketing automation and offer customization. In partnership with the supplier of timekeeping equipment, we developed a solution that allowed users to send their timings and ratings soon after crossing the finish line.
After test reading QR Codes printed on shirts and the OCR of numbers in the athletes' chests, we created a tool to process and identify event photos. For identification of chest numbers, 3 computer vision APIs were analyzed: Google Cloud Vision, IBM Watson and Amazon Rekognition. Following evaluation of factors such as error rate, processing time and recognition cost, the Google API was chosen for having the lowest error rate, lowest processing time and lowest percentage of false positives.
Together with Track&Field, we were able to increase the application's user base, promote new events and attain high participation rates.
active trainers on the platform