Wrap-Up: Brand Safety Week and What's Coming for Audio Adtech
By Bo Carper
Last week I attended the Brand Safety Summit in NY, where industry experts and senior leaders in the media industry came together for timely conversations that promote safety and suitability for brands and consumers in digital environments. It was great to see familiar faces and make new connections during the day-long sessions.
In thinking about the future of advertising as it applies to podcasting and the spoken word, here are the four themes and highlights that I found to be the most interesting and relevant from the event:
#1 CTV performance & quality expectations are highly relevant to podcasting
So many similarities exist between podcasting and CTV, especially as more and more podcasts are released with both audio and video streams. Marketers expect the same quality and brand suitability performance from CTV and podcasting that they do across their other media channels.
Deva Bronson from Dentsu was joined by representatives from DoubleVerify and Snap in a conversation centered around maintaining quality for CTV, especially as more advertisers use programmatic pipes for ease of targeting and measurement. The auction environment is bringing in great opportunity and increasing the urgency for world-class tools and technology to classify such content in near-real time.
#2 Algorithms are failing multicultural communities.
As AI & machine learning become more prevalent across the ad tech ecosystem, it is clear that we risk suppressing multicultural voices because of unconscious algorithmic bias that creeps into machine learning models.
"Preserving the Democratic Internet: Avoiding Inherent Bias in Digital” was a great session on this topic, featuring Alyssa Boyle from AdExchanger, Karan Dalal from Media.net, Wendi Dunlap from Kinesso & Betsy McLeod from H&R Block. This session made me proud of the research we’re doing at Sounder to constantly improve our models and look for algorithmic bias so that we are part of the solution and not the problem.
#3 Combining image recognition & brand safety with audio brand safety is going to become more and more necessary.
Deep fakes are going to present unique challenges. “Brand Suitability in the Age of Misinformation” was a great session that highlighted just how difficult it can be to tell the difference between a faked photo and an image. It was both enlightening and concerning to be presented with various images, one real and one fake, and be completely unable to discern a difference with the naked eye. Eventually, we’ll need to address the same problem for the naked ear.
#4 Spotify is raising the bar on brand safety.
My former colleague (although we only overlapped for a month) David Byrne highlighted Spotify’s efforts to establish a high bar for brand suitability. He gave a great presentation which really illustrated the need for nuance in language understanding, by showing how the same English language word can be perceived very differently (and be less brand suitable) depending on which English speaking country one is residing in.