Finding the Sweet Spot: Ad Scheduling on Streaming Media
Prashant Rajaram, Puneet Manchanda, and Eric Schwartz, 2020, 20-116
Most U.S. households subscribe to at least one video streaming platform. Viewers consume content on streaming platforms in a self-directed manner since these platforms, in contrast to live TV, are primarily on-demand services.
On the platform side, the tracking of individual-level viewer “eyeballs” on streaming platforms represents an attractive opportunity for advertisers, especially as these services allow for ad personalization based on the availability of rich data. At the same time, however, interruptions to the viewing experience via advertising can detract from the viewers’ feeling of being in control, which may lead to decreased content consumption.
Past research in marketing does not provide much guidance for developing an optimal advertising schedule that does not detract from the viewing experience. Research has typically focused on ad-scheduling in live TV settings where the viewer has very limited control in terms of how and when she consumes content. Moreover, this work has optimized outcomes such as sales (as a function of ad exposure) but has not studied the effect of ad-schedules on viewers’ consumption patterns.
Here, Prashant Rajaram, Puneet Manchanda, and Eric Schwartz develop a three-stage approach to deliver an optimal advertising schedule that maximizes advertising exposure without compromising the content consumption experience for individual viewers. Their approach is calibrated on a unique dataset that tracks the viewing behavior of Hulu customers.
In the first stage, they develop two metrics - Bingeability and Ad Tolerance - to capture the interplay between content consumption and ad exposure in streaming media settings. Bingeability represents the number of completely viewed unique episodes of a show while Ad Tolerance represents the willingness of a viewer to watch ads and subsequent content.
They use summary features from the current and past viewing environment to predict the value of the metrics for the next viewing session. Simultaneously, they control for the non-randomness in ad exposure via the use of ad exposure patterns of other viewers watching the same content in the past. By leveraging causal machine learning and artificial intelligence methods, they are able to do this at the viewer-show-episode level.
Finally, they use the output from the previous stage to design an “optimal” ad schedule for a given viewer, making sure that it does not exceed her Ad Tolerance, and develop this into a full decision support system, allowing the platform to explore ad schedules that tradeoff content consumption and ad exposure for its viewers.
Put into Practice
This study provides an approach for streaming providers to explore the tradeoff between content consumption and ad exposure in order to provide a balanced viewing experience. Interestingly, the authors show that “win-win” ad schedules that simultaneously allow for higher content consumption at higher levels of ad exposure are feasible. Further, their decision support system can be integrated into an online experimentation platform, where recommendations can be tested in live settings and results from experiments can be used to improve the performance of predictive models.
- Corporate: FREE
- Academic: FREE
- Subscribers: FREE
- Public: $18.00
3 WAYS to GET CONNECTED
Employees of MSI Member Companies enjoy the benefits of complete online access to content, member conferences and networking with the MSI community.
Qualified academics benefit from a relationship with MSI through access to msi.org, conferences and research opportunities.
The public is invited to enjoy partial access to msi.org content, a free e-newsletter, selected reports and more.