To improve the relevance of the offers, and provide better insights to the clients, we needed to improve the attribution matching.
Rokt is a technology company specializing in e-commerce marketing. It provides personalized transaction moments to enhance customer engagement and revenue. As of 2023, Rokt has over 500 million users and is valued at approximately $2.4 billion. Rokt has partnerships with notable companies such as Uber, Ticketmaster, Fanatics, and Live Nation.
I worked at Rokt as a Senior Data Engineer for five years. My work involved developing services in a high-performance, high-availability, scalable platform that processes billions of monthly transactions to deliver personalized and relevant offers to millions of users worldwide.
Properly attributing the campaign events to the desired outcome is a complex task that requires comparing many events and applying different attribution rules per campaign to ensure consistency and explainability. Properly measuring how successful a campaign is is crucial for clients to understand the return on investment. It is also important to guide the machine learning models in better predicting and optimizing the campaigns.
The new attribution system generates the matching events in a fraction of the time and with a fraction of the resources. It also provides new features to the clients, such as better explainability, monitoring new events during the customer journey, consuming new nurturing events, and changing the attribution rules on the fly. By using the new events, the machine learning models could learn and optimize the campaigns faster, improving the return on investment.