The podcasting advertising industry looks a lot different today than it did even two years ago thanks to programmatic audio ads.
While the bread and butter host-read ad format still packs a punch when it comes to driving listener engagement and conversions, those ads aren’t scalable, said Lizzie Widhelm, svp and head of ad innovation and B2B marketing at SiriusXM Media. And in 2023’s slow ad market, the ability to target large swaths of audiences based on their demographics and interests rather than a specific show is what helped SXM Media maintain positive revenue momentum in 2023.
As of November, SXM Media has “already exceeded our [revenue] goal in podcasting,” said Widhelm. “We’re doing more revenue now – dramatically more – than last year. We have almost a 90% increase in programmatic revenue specific to podcasting … So it’s exciting to be where we are at today, but there is still road to go.”
In the latest episode of the Digiday Podcast, Widhelm chats about the advantages of selling audio ads programmatically and what the next stages are for improving audience targeting capabilities in the medium, as well as how generative AI can be used to automate the programmatic audio ad sales process even further.
Below are highlights from the conversation, which have been lightly edited and condensed for clarity.
The growth of programmatic audio ads
Over the course of the last two years [programmatic became fully implemented]. It started with first saying, what is the currency? Is it downloads? Is it impressions? Well, if it’s going to be impressions, we need to dynamically insert the creative for a number of reasons to be able to count properly, to be able to start and stop creative to be able to swap creative for the most timely message … And so we worked really hard, I’d say two years ago, to just establish all of that.
Where there’s still innovation to come, is on the future of just all the audience segmentation work. So we have onboarded partnerships with Comscore and others, of course we have strong first-party data, but we’re a distributed strategy … And so I’m just getting the future of like, how can you not just buy programmatically into a group of shows or into a single show, but I want to buy moms across the country. And I want to have this more layered, more specific-to-streaming playbook in podcasting. And we’re well on our way there.
Scale isn’t everything though
We’re working so far upstream at the strategy layer with many of these brands that oftentimes we will recommend a custom program [like] do a sponsored segment with a host that you think really aligns to you. Yeah, you’re not going to get the same reach, unless you’re picking a host that has a really wide distribution, but the power of our podcasting hosts [or] our live hosts on air over at Sirius … we can find them the right way to integrate in the audio space that meets their needs.
And it’s OK if they’re not at a place yet where the world of podcasting at scale is right for them. There’s still so much else to buy … if you want to reach 25- to 50-year-old women in podcasting. Partner with Tinx, partner with Radio Andy, partner with Nicole Byers. There’s so much great content.
Streaming ad production with AI voice cloning
I think generative AI for creative production [can] really improve the time spent for creative production. We’ll see voice cloning definitely be a place we can play.
Back to your question about getting creative approvals, giving spec spots to advertisers to get them more comfortable [or] small business advertisers that don’t need a host. And so you’re just building effortlessly, building creative really fast. So I think there’s going to be a lot of testing and learning there … to try to shore up the gap from point of purchase to live campaign. I just don’t think we’ll ever be putting fake Conan on the air.
[Maybe] when Conan is on vacation, we can have his clone voice record a bunch of spots to get them approved so when he comes back from vacation, he’s ready to rock. You can obviously see how that is a big lift to [the] production cycle.