Written by:
Stay updated with Roku Advertising.
Beth Logan doesn’t believe that artificial intelligence (AI) will take over the world. But Logan, Roku’s VP of Data Science and Voice, does see AI-powered tools and chatbots someday becoming as ubiquitous as calculators and spreadsheets.

Logan, an Australian native with computer science and engineering degrees and a PhD in speech recognition, previously served as dataxu’s SVP and Chief Data Scientist. Today, she sits at the forefront of Roku’s efforts to use AI to improve the user experience for its user base of 70 million active accounts. Her team is also leveraging AI to gain insights and help Roku’s advertisers meet their goals.
We sat down with Logan to discuss how Roku is using AI, the biggest opportunities for advertisers, and how AI may address the advertising industry’s greatest unsolved challenges.

We’re seeing intense interest in AI. How would you define it?
The definition I like is the capability of computers to imitate intelligent human behavior. Speech or image recognition are obvious [use cases], but we can also apply it to things we do in advertising like real-time decision-making or making forecasts. We're always predicting how much an ad will be worth, who's going to react to it, and if consumers provide value back to the brand by purchasing a product.
What are the most exciting opportunities for AI in advertising?
I think television commerce (T-commerce) is very exciting for Roku and advertising, and there are definitely opportunities for AI there.
And we've merely scratched the surface of causality in advertising. Many times, we measure what we see: We saw these ads and this many people converted. But did the ad persuade people? Because that is really worth spending money on, versus situations where ads lead to a lot of conversions, but where those viewers would have converted anyway.
What types of problems is your Data Science team trying to solve using AI?
We're looking at what sort of ads to show, and aspects of that problem like bidding for ads on the ad exchange, pacing and user experience. When dataxu merged with Roku, I started applying data science to other areas, like creating models for customer lifetime value (CLV). We also work on anomaly detection, so that when we roll out different firmware releases, we can be smart about detecting problems.
Can you share a few key use cases for AI at Roku?
At Roku, we use AI to help users find content to watch. When you turn on a Roku TV or a Roku streaming player, there's a user interface that is largely driven by machine learning to bring content to users that's interesting to keep them engaged.
On the advertising side, we have a feature called Maximize Actions, which is when you're looking for new customers. We had implemented it previously at dataxu for banner ads and mobile, and when we joined Roku, we had more TV streaming data, so we implemented it for the TV streaming use case. It finds users who will convert for a particular KPI for different products, and it’s one of our most popular capabilities.
We also have a new feature called Maximize Reach, with the option to set a target frequency. Advertisers want to reach as many people in their target audience as possible, and then serve each person, say, five impressions. Previously, they had to juggle things to make that happen, but we have a new algorithm that will automatically spread out the spend across the audience at your desired target frequency. What can happen is one household may see too many impressions, which doesn't leave much budget left over for other households. You'd rather spread it out so that everyone gets a chance to know about the product when you're reaching an audience.
What unsolved advertising challenges could AI address in the future?

If the ads are correctly reached, people don't need to watch as many because they'd be more effective. Everyone would be happy. I think that's an opportunity for the future.
What are you currently streaming?
I’m streaming murder mysteries on BritBox – currently watching “Shetland.” And because I love weather and live in Boston, I'm a big fan of WeatherNation on The Roku Channel.
Subscribe to our newsletter to learn more about Roku Advertising.
