by Gavin Gaddis

Not to brag or anything, but it’s getting cold enough to necessitate house slippers and I’m having soup for dinner. It’s the best time of the year in the hilly corn fields of Indiana. Before I dig out my flannel button-ups and denim jacket, let’s look back at this week’s developments in the podcasting world.

Tom Webster writes from Content Marketing World in Denver, where the event shared an expo hall and keynote stage with LIONS and The Market Research Event. Putting marketers, creatives and researchers under one roof should have produced three conversations, but nearly every booth was selling an AI-augmented product, so many that he could not tell which conference he was standing in. Webster takes umbrage with the hype, not the technology. A large language model generates text by predicting the most probable next word, which makes it a machine for producing the most likely, and therefore most average, answer. The latest models can do the job of the average content marketer, and he sees the same risk in market research, where synthetic respondents were the hot topic. Research earns its keep by finding the outlier nobody modeled, and a synthetic respondent simply hands the consensus back to you.

For podcasting, Webster says the dividing line is the same: AI used for average tasks makes you better, and AI used to make the product itself makes you average. Show notes, chapter markers, and ad operations such as trafficking, reporting and matching inventory to briefs are all fair game to delegate.

What can’t be delegated is the work predicated on the individuals’ exceptional traits. The host-read ad, and the content itself. The host read works because the host is a trusted figure the audience believes is vouching for something. A synthetic voice reading generated copy can only create the most average, probable version of an ad that listeners have trained themselves to skip.

His prescription is to reset your internal thermostat. That internal point for what counts as “good enough.” If yours sits at “good enough,” you are competing with a machine whose entire job is “good enough.” Handing that machine the average work allows one to reinvest their hours in what it cannot do.

Tom Webster will be back in Avengers: Doomsday with another presentation next Tuesday, October 13, at 2:00 p.m. ET, co-hosting a webinar with ADOPTER Media founder and CEO Glenn Rubenstein. They will present a new Podcast Atlas analysis of podcasting’s heaviest buyers, The Super Spenders: Who Really Buys From Podcast Ads.

In a guest article for Sounds Profitable, Audacy Head of Podcasts Leah Reis-Dennis argues that much of the advice brands still follow was built in podcasting’s earlier, scrappier years. She notes that advertisers are pouring in, with monthly spend per new advertiser more than doubling in the last year and ROI now stronger than social, display, video or search, according to eMarketer and Magellan AI. Her first target is the assumption that a bigger audience means bigger impact. Podscribe data shows advertisers on mid-sized shows see a 22% lower cost per acquisition than those on the largest celebrity-led podcasts, and an Audacy study of 1,000 U.S. adults found listeners are 21% more likely to be influenced by hosts with real topic expertise than by people with large social followings. ADOPTER Media’s Adam McNeil says his team tends to stay away from macro podcasts because a celebrity’s connection to the audience weakens as their status grows.

On format, Reis-Dennis says the fear that video would cannibalize audio has largely not borne out. Edison Research data shows 57% of Americans both watch and listen to podcasts, and 72% of podcast audio listeners say they started with a show’s video version, which she frames as video driving discovery while audio drives deeper, screen-free engagement. Mynt’s Christine Hess calls podcasts a hybrid video and audio buy, with video bringing visual ad real estate audio never had. AI is where she draws the line. Weekly podcast consumers accept it for back-end work like editing, research and transcription, but just 25% approve of AI-hosted shows and 62% see AI as at least somewhat of a threat to podcast credibility. Oxford Road’s Dan Granger warns that AI could erode the host-audience trust the industry is built on.

That trust is the variable the old playbook never measured. eMarketer data shows 75% of consumers rank podcast hosts among the most authentic voices recommending a product, compared with 54% for social media influencers, and Audacy research finds 76% of podcast listeners have purchased a product after hearing an ad from a podcast creator. Her conclusion is that the best opportunities for buyers are shows built on deep connection, whether a mid-sized audio show with a loyal audience or a massive video podcast whose host carries trust across formats. The full research is in Audacy’s State of Audio report, The New Rules of Podcast Advertising.

The second report from Signal Hill Insights’ Audio on the Move study, Canada’s first holistic look at audio listening, draws on a nationally representative survey of 6,161 Canadians aged 12 and older. Respondents reported on their listening “yesterday,” with weekdays and weekends sampled proportionately and results balanced for age, gender, education, language and region. AM/FM radio accounts for 46% of in-car listening time, followed by music streaming services at 27% and music on YouTube at 11%. The car is where 24% of Canadians’ audio time happens, compared with 59% at home and 12% at work or school. Signal Hill’s Matt Hird says streaming is claiming a growing share, and that combining YouTube and music streaming closes the gap with AM/FM quickly. The first report had already found music on YouTube holds the second highest share of ad-supported audio overall.

The study also tested attitudes toward content fully created with AI, a question that lands in a country the Ipsos AI Monitor 2026 ranks last of 32 for excitement over AI products and first for nervousness about them. Only 4% of Canadians like AI-generated podcasts and 5% like AI-generated music, while 68% and 65% dislike them, with 28% and 29% neutral. Hird puts the gap at a factor of at least 13 to 1. Canadians aged 12 to 17 are less negative, with half disliking either format and 42% neutral on AI-generated podcasts, but sentiment still does not tip positive. For context, 60% dislike AI-generated images or memes on social media, the fewest of any format tested.

WARC Media forecasts that global ad spend will grow 11.9% to $1.34 trillion in 2026, following 10.0% growth in both 2024 and 2025. It credits significant corporate AI investment and major events including the Olympics, the FIFA World Cup and the US midterm elections. WARC’s Suzy Young calls these unusual times, with investment accelerating even as many consumers face cost-of-living pressure, a contradiction she attributes to an increasingly uneven economy in which the AI boom benefits some companies, sectors and consumers more than others. Further escalation of global tensions is the main downside risk.

Growth is expected to moderate to 8.4% in 2027, reaching $1.46 trillion, and to 7.9% in 2028, reaching $1.57 trillion, which would make the market 2.3 times larger than it was in 2019. WARC also says AI assistants and generative search are becoming new places for brands to advertise, as ad dollars follow product discovery and purchasing to those gateways.

For the first time, Podscribe’s Rising Advertisers Ranker is now available to the public!

The ranker highlights advertisers that have increased their podcast ad spend by more than 50% month over month, ranked by estimated ad spend.

It’s an easy way to spot which brands are quickly investing more in podcasting—and where advertiser momentum is building.

👀 See who’s on the rise in the September rankings → 

We’ve also made two major updates to our podcast industry rankings: 

While they may not be top story material, the articles below from this week are definitely worth your time: