
Is artificial intelligence better for ad targeting than first-party data? Marketers will now be able to put it to the test as Podchaser has launched an industry-first predictive language modeling capability that allows advertisers to refine their podcast audience targeting. By using AI, Podchaser says it is able to analyze the language spoken within a podcast to predict the age and gender of its likely audience.
“In bringing Predictive Demographics to market, we modeled and compared our predictions against trusted demographics to ensure accuracy,” CEO Bradley Davis says. “While still in the early stages, Predictive Demographics has seen a high success rate and proven to be among the most accurate models for demographic data and insights. As the AI powering the tool continues to learn from more podcasts the success rate continues to improve.”
Buying against demos like age and gender is nothing new, of course, but as ad cookies have faced an expiration date, more marketers have begun to use first-party data to collect these insights from media companies including podcasters. But Podchaser sees an opening for AI pointing to a Nielsen report that shows 56% of brands surveyed said first-party data is “below average,” “average,” or “average at best” when it comes to actually being put to use.
Podchaser’s Predictive Demographics has limited demographic targeting capabilities at launch, but that is expected to change going forward as the company further trains its algorithms and as it expands its Collections+ product of which Predictive Demographics is a feature. Launched in June, Collections+ use AI to compile data on podcasts and their audiences to allow marketers to target podcast listeners using a collection of data inputs.
“As a data-driven company, Podchaser has and continues to analyze these predictive AI models,” Davis says. “We also know that AI is most valuable when the people behind it are constantly examining the results to help the algorithms better learn and achieve the goals as well as to inform their own strategies for using the technology.”
As marketers have looked at their options in the post-cookie world, several podcast companies have been presenting alternatives that have leaned on the first-party data that they collect. Libsyn last month unveiled its new Predictive Contextual Targeting that leverages audience data, advanced speech-to-text transcription, and intelligent categorization. And last year SXM Media expanded its relationship with Comscore to create Predictive Audience Targeting that, when combined with AI transcription technology, extracts the contextual essence of the content allowing brands to buy against it.
Davis says predictive demographics is also one of many models that Podchaser uses to arm its customers with the best insights to fit their needs.
“While some of these models do follow a more ‘data-based’ style, we are fully aware of the weaknesses that data-based approaches can have – notably sampling bias or low response rates from surveys or the privacy issues surrounding surveillance-based systems,” he says. “Predictive Demographics is our solution to these problems and improve the accuracy of the insights that we provide our customers.”
Ultimately, Davis sees the use of AI as a way to grow revenue for the industry as podcasters are able to offer more comprehensive targeting for advertisers than ever before, including “smart” demographic breakouts.
“We’re making it possible for advertisers to unlock value from podcasts through an audience-first approach and to ultimately discover podcasts with untapped advertising potential,” he says. “This in turn helps more podcasters generate more revenue – a win for all.”
Since Podchasher debuted Collections+ three months ago, its parent company Acast has utilized the technology to power campaigns for over 500 clients across 14 markets. It says it has also allowed it to monetize 10% more shows.
Comments