How Pandora turns personalization into $1 billion in ad revenue | Industry
Breaking Tech Industry news from the top sources
Personalized web and mobile ads that reflect your customers’ tastes, values, and preferences is the key to keeping them clicking, but it’s hard to pull off. To learn how companies like Pandora flawlessly serve up customized advertising that always clicks, don’t miss this VB Live event!
Register here for free.
Personalization is the core of Pandora, says David Hardtke, director of advertising science at internet radio station Pandora, which has more than $1 billion in digital advertising revenue.
His team focuses on optimizing the Pandora platform on behalf of their advertisers, using music listening preferences and music listening data to get insights into who their users are. The goal is determining when and where to play ads, using intelligent ad search technology powered by AI and machine learning, as well as improving measurement on whether or not an ad was heard and whether it was effective for the advertiser.
In Pandora, users start a station, seeded by an artist or a song or a genre. As Pandora serves up related music, the listener provides feedback in the form of thumbs up or thumbs down for the songs they like and don’t like. From there, Pandora builds a personalized station based on that user’s feedback, as well as the feedback from other listeners on that feed.
And on the advertising side, it turns out that the music you listen to is one of the more powerful signals into getting an insight into who a customer is, Hardtke says.
“The average Pandora listener listens 24 hours a month, so we have this wealth of personal data: when you listen, what devices you listen on, and what you like to listen to,” he explains.
Pandora’s raison d’être is building algorithms that personalize stations for their users, and their data science team is focused on ensuring that the music is right for each individual listener — and they then can apply that to their advertising. They can build custom targeting segments based on listening behavior and other behaviors.
For instance, Pandora can figure out when you are listening with your kids and which kid you are listening with. They can identify when you’re in the car with your kids, and when you drop them off at school. They can figure out when you’re working out, and surface not just workout-appropriate music, but also workout-appropriate ads, and then build tests with advertisers.
“The powerful thing is that we find these patterns of listener behavior and interactions identify really specific, really niche groups of people,” Hardtke says. “We’re only just now starting to exploit that and help our advertisers use that data in their campaigns. We see the immense power to hitting the right audience with their message.”
Hardke warns that it can be overdone, completely skewing your data and serving up advertisements that are not just incompatible, but insulting to your users, such as when you can engage in what’s called overfitting, which is over-personalization based on a limited set of data. An example of overfitting would be during the American election, in which the Rasmussen poll consistently showed Trump having a larger than expected vote share among black Americans than any other poll.
He also points out that there’s a very fine line between hyper-personalization and trying to have a broader reach. When you’re trying to narrowly target the people you think are in the market, you’re missing out on people who have unanticipated future needs — but hyper-personalization still unlocks a world of opportunity for brands that want to keep themselves top-of-mind, such as the listener realizing their insurance is about to inspire. And who could turn down the possibilities of personalization that follow your customer throughout their journey, from mobile to web and all during the day?
To learn more about how Pandora transforms rich user data into advertising dollars, how to unlock the potential of hyper-personalization, and how to avoid the biggest mistakes, don’t miss this VB Live event!
Don’t miss out!
Register here for free.
Attend this webinar and learn:
- How to offer your customers content that connects to their lifestyle and tastes
- The intersection between art and science in the new world of hyper-personalized advertising
- How to create customized content that connects without creeping out customers
- Best practices companies like Pandora and Trulia use to create authenticity and bring in more revenue
- David Hardtke, Director of Advertising Science, Pandora
- Deep Varma, VP of Engineering Trulia
- Rachael Brownell, Moderator, VentureBeat
Sponsored by Cvent