The computational demand in digital advertising has dramatically increased in recent years. EMarketer has forecasted that almost $19 billion in additional ad spending will enter the programmatic space between 2018 and 2020. By 2020, nearly nine in 10 mobile display ad dollars will transact programmatically.
While programmatic advertising enables both buyers and sellers to scale their campaigns more efficiently, it also requires extensive data processing and computational costs. To address these issues and cut down on costs, the ad tech industry has promoted the application of machine learning (ML) techniques. Although the use of ML applications is still nascent in the ad tech industry, the vast availability of data is an intriguing opportunity for data scientists looking to evolve the industry. Collaborative investigation and knowledge-sharing will be crucial to the successful application of ML principles to optimize programmatic advertising.
Recently, I attended KDD 2018 in London, a knowledge discovery and data mining conference that has been held annually since 1995. The fundamental focus of KDD is to bring players from a variety of disciplines together from all around the world to concentrate on developing techniques and methodologies in data science, machine learning and data mining with the goal of addressing future industry challenges. This was a perfect knowledge-sharing forum for the Motive Intelligence Team, which has drastically transformed Motive’s native bidding strategies into programmatic knowledge over the last few years.
KDD 2018 hosted more than 3,000 ML experts from industries ranging from healthcare to advertising to present data that addressed the following key questions:
- Is the industry ready for a change?
- How can ML methodologies be leveraged by their business?
- How can ML methodologies be scaled?
- What are the current and future challenges posed by ML?
Along with addressing these questions across different industries, KDD 2018 exhibited an interdisciplinary rally, where one might find oneself in discussion about the application of the ‘Lagrange’ equation to target Snapchat users in one second and then a panel weighing the revenue and cost tradeoffs of quantum machines in the next. Participation was highly diverse as well, with top tech firms like Amazon, Google, Microsoft, IBM, Alibaba providing benchmarks for scale and platform levels, and research labs from prestigious universities providing innovative learning algorithms to pave the way for future innovation in ML. In this way, KDD was the optimal intersection of industry techniques and theoretical insights to address the opportunities posed by machine learning in the data science world.
The KDD conference really opened my eyes to the both the challenges, complexities and, most importantly, great opportunities for companies that can successfully apply ML techniques to the ad tech industry. Motive has made significant improvements to its programmatic platform with campaign automation, smart bidding and performance pacing tools, all with the ultimate goal of helping our clients achieve mobile user acquisition performance at scale. While we have more to do, the opportunity to apply ML is very exciting for us.
Overall the conference was a great experience and opportunity to network with brilliant minds are trained by similar phenomena: the science of data. Special thanks to all the organizers, participants and volunteers who helped curate and present the high quality of insights at KDD this year!
