We listened to 113 hours of podcasts and counted every second of ads

128 episodes. 107 different shows. Every segment timestamped and classified. We built a pipeline to transcribe podcast audio and detect advertising, then ran it on a sample from across the podcast world. The question was simple enough: how much of a podcast episode is actually ads?

The answer depends wildly on what you listen to.

Why bother?

Podcast advertising is a $4 billion industry, and growing. But as listeners, we don't really think about how much time we spend on sponsor reads, cross-promos, and Patreon pitches. We hear "let's take a quick break" a dozen times a week and just accept it.

We wanted real numbers. Not industry self-reports, not estimates. Actual measurements from real episodes downloaded and analyzed in early 2026.

How we did it

We used OpenAI's Whisper through the faster-whisper library to transcribe each episode with segment-level timestamps, saving everything as structured JSON. Without those timestamps, you can't calculate what percentage of an episode is advertising.

For ad detection, we fed each timestamped transcript to Claude, which classified every segment as content, sponsor ad, cross-promotion, or listener support. AI analysis was the right tool here because advertising in podcasts is surprisingly hard to identify by surface features alone. Host-read ads don't always sound like ads. A host might spend 90 seconds talking about how much they love a mattress, mention a URL at the end, and never once say "sponsored by." A Swedish pre-roll can show up in an English podcast. A guest discussing their book as part of a genuine interview looks nothing like a scripted sponsor spot, even though both mention a product. You need something that understands context and intent.

With 128 episodes to process, we ran transcription as a background batch job and launched analysis agents in parallel. The whole thing, raw audio to finished JSON reports, ran on its own.

The numbers

Here's what 113 hours of podcast audio breaks down to:

Metric	Value
Episodes analyzed	128
Unique podcasts	107
Total audio	113.1 hours
Total ad time	6.15 hours
Average ad percentage	5.44%
Median ad percentage	3.73%

The average episode is about 5.4% ads. The median is lower at 3.7%, which means a handful of ad-heavy shows are pulling the average up. That gap between mean and median tells you the distribution is skewed right: most shows are fairly restrained, but the ones that aren't really aren't.

In practical terms: a typical 60-minute podcast episode has about 2 minutes and 15 seconds of ads. For comparison, a 60-minute slot on commercial radio is roughly 15-20 minutes of ads, and an hour of ad-supported streaming video runs about 4-8 minutes. Podcasts are still the lightest ad experience in audio or video media.

Where the ad time goes

We broke ads into three categories:

Ad type	Total hours	Share of ad time
Sponsor ads (paid)	3.38 hours	55%
Cross-promotions (self/network)	1.37 hours	22%
Listener support (Patreon, donations)	1.05 hours	17%

Paid sponsorship is still the big one, more than half of all ad time. But the second line is interesting. Cross-promotions, hosts plugging their own books, courses, other shows, or network siblings, take up nearly a quarter of all ad time. These aren't paid placements in the traditional sense. They're a show using its own airtime to drive audience to related properties. From a listener's perspective the experience is the same as an ad, but the economics behind it are different. A network cross-promo costs the show nothing to run and builds audience for sibling shows. It's free inventory, so networks use a lot of it.

Patreon and donation pitches account for 17%. That's worth noting because listener-supported shows often describe themselves as "ad-free." Technically, they have no paid sponsors. But the Patreon pitch itself takes time, and from a listening experience standpoint, it functions the same way: the content stops, the ask begins. Whether you find that more or less annoying than a BetterHelp read is a matter of personal preference.

The spread is enormous

This is where the averages stop being useful:

Ad load	Episodes	Share
0% (completely ad-free)	17	13%
Under 2%	22	17%
2-5%	42	33%
5-10%	23	18%
Over 10%	24	19%

A third of episodes land in the 2-5% range. These are shows with one or two quick sponsor reads, then back to the content. You probably don't even notice the ads. Combined with the ad-free and under-2% groups, that's 63% of episodes at 5% or less. The majority of podcasts treat advertising as a minor interruption.

But nearly one in five episodes crosses 10%. At that level, a 60-minute show has more than 6 minutes of advertising. And some go way beyond that. The top of our sample hit 27.6%, meaning more than a quarter of the episode was promotional content.

That spread, from zero to 27%, within a single medium, is remarkable. You wouldn't see that range in television, radio, or streaming video. Podcasting has no FCC guidelines on ad minutes per hour, no platform-enforced limits, no industry standard. Each show sets its own line. Some set it at zero. Others set it wherever the money takes them.

The heaviest and lightest

The five episodes with the most ads:

1. Good Moms Bad Choices (iHeart) at 27.6%. The same bundle of five network cross-promos played four separate times. Four times. The same ads. This is the iHeart model at its most aggressive: take a block of network promos and insert it at every break point. The actual show content was fine. The surrounding ad scaffolding was relentless.
2. The Birth Hour #1036 at 24.8%. A 60-minute episode with an 11-minute "interview" with a sponsor rep tacked on after the real content ended.
3. The Birth Hour #1037 at 24.7%. Same show, same format. This time a 13.5-minute sponsored interview. More on this pattern below.
4. Tumble (The Science of Fungi) at 24.6%. A kids' science podcast. It front-loaded a 97-second self-promo, ran multiple Patreon segments, and hid a promo code in an "Easter egg" at the end. Creative, but it adds up.
5. Not Your Average Runner at 21.6%. A short episode (27 minutes) with a 3.5-minute sports bra sponsor read. The short runtime inflates the percentage. In absolute terms, 3.5 minutes of ads isn't extreme, but against a 27-minute episode it looks heavy.

On the other end: seventeen episodes had zero ads. Church sermons, gaming roundtables, academic interviews, niche hobby shows. A Mindscape philosophy episode ran 78 minutes without a single promotional mention. Two Rebel FM gaming episodes went over an hour each, completely clean. A children's bedtime story show. Several independent podcasts that just don't bother with ads at all.

There's a pattern in that list. The ad-free shows tend to be passion projects, institutional content (churches, universities), or community-driven shows where the host has no financial pressure to monetize. They exist because someone wanted to make them, not because someone needed to make money from them. That's a luxury not every podcaster has, but it tells you something about where the cleanest listening experiences come from.

The network effect

iHeart shows were the heaviest in our sample, and the reason is structural. Their model relies on inserting bundles of cross-promotions for other iHeart shows, and they repeat the same bundle at multiple break points. The Good Moms Bad Choices episode played the same five-promo block four times. That's 20 cross-promo slots in a single episode, all for shows on the same network.

This isn't unique to iHeart. Network-owned shows across our sample had higher ad loads than independent ones. Networks have the infrastructure for dynamic ad insertion, relationships with advertisers, and an incentive to cross-pollinate their audience. An independent show with 5,000 listeners might not bother with ads at all. A network show with 50,000 listeners has a sales team filling inventory.

The practical takeaway: if a show is part of a large network (iHeart, Wondery, Spotify/Gimlet, NPR), expect more ads. If it's independently produced, expect fewer. This isn't a hard rule, some network shows are very light and some indie shows plug their merch relentlessly, but as a general pattern it held up across our 128 episodes.

Dynamic ad insertion and the Swedish ads

Something we didn't expect: Swedish ads kept showing up in English-language podcasts. BBC shows, American history pods, fantasy baseball, a leadership coaching show. All of them had dynamically inserted pre-roll ads in Swedish for things like AJ Produkter (office furniture) and McDonald's.

It makes sense once you think about it. We downloaded the episodes from Sweden, and the ad servers geo-targeted accordingly. But it highlights something important about modern podcast advertising: the ads you hear aren't necessarily the ads someone else hears. Dynamic ad insertion means the ad experience is personalized by geography, listener profile, and timing. Two people downloading the same episode on the same day in different countries will hear different ads, and potentially different amounts of advertising.

This makes any study of podcast ads inherently local. Our numbers reflect what a listener in Sweden heard in early 2026. Someone in the US, UK, or Australia downloading the same episodes would get different pre-rolls, potentially different mid-rolls, and a different total ad load. Industry reports that claim "the average podcast is X% ads" rarely acknowledge this. The average depends on where you are.

Native advertising and the blurred line

The Birth Hour gave us the clearest example of native advertising in our sample. Both episodes followed the same format: 45-50 minutes of genuine interview content, then a 10-15 minute segment presented as a "bonus interview" with someone from the episode's sponsor company. The transition was smooth. The tone was conversational. If you weren't paying attention, you might think the show just had a second guest.

This format is clever because it doesn't feel like an ad break. There's no "and now a word from our sponsors." The sponsored interview is presented as additional content that happens to feature someone with a product to sell. It pushed both episodes past 24%, but a casual listener might not register that a quarter of what they heard was paid promotion.

We saw softer versions of this across other shows. Hosts weaving product mentions into their regular commentary. Guests who happen to have a book or course coming out, with the host helpfully providing the URL. The line between content and promotion has always been fuzzy in podcasting, and it's getting fuzzier. Unlike a 30-second pre-roll that's obviously an ad, these integrated mentions resist easy classification. We did our best to draw consistent lines, but reasonable people would disagree on specific calls.

Sleep podcasts, vintage ads, and other oddities

Sleep podcasts keep their ads short and front-loaded. Which makes sense. Nobody wants to get startled awake at the 30-minute mark by a mattress commercial. The shows in our sample that were designed for falling asleep to clustered all their promotional content in the first two minutes and then ran uninterrupted for the rest. Smart design for the format.

One of the more unusual finds: "Down These Mean Streets" rebroadcasts classic 1940s and 50s radio dramas, complete with the original commercial breaks. Ex-Lax. Colgate shave cream. Ford Motor Company. Vintage ads preserved as historical artifacts. They're charming in a way that modern ads aren't, but they still take up airtime. We counted them. An argument could be made that they're content rather than advertising at this point. We went with "advertising" because they were literally produced as ads, even if they're now 75 years old.

Tumble, the kids' science podcast, gets points for creativity. The show hid a promo code inside an "Easter egg" segment, turning the ad into a game for young listeners. It's a thoughtful approach to a hard problem: how do you run ads on a show for children without alienating parents? The answer, apparently, is to make the ads part of the experience. It still pushed the episode to 24.6%, though.

The listener-supported difference

Listener-supported shows really are different, and the data backs this up clearly. Effectively Wild, the FanGraphs baseball podcast, ran 96 minutes at 0.2% ads. Jay and Miles Explain the X-Men came in at 2.2%. Boring Books for Bedtime, which explicitly bills itself as "100% listener-supported and ad-free," was at 2.8%, just the Patreon thank-you segment.

That 2-3% for the Patreon pitch is interesting. It's the cost of being listener-supported: you still have to ask for money. You just ask for it from your audience instead of from advertisers. The pitch is shorter, less frequent, and more personal than a scripted sponsor read, but it exists. Truly zero-promotion podcasts are rare. Even the ones that don't need money tend to plug their social media or ask for reviews.

The tradeoff between paying $5/month and skipping the ads appears to be real. If you're a heavy podcast listener doing 2-3 hours a day, and you listen to ad-supported shows averaging 5%, that's 6-9 minutes of ads daily, or about 40-60 hours per year. Whether that's worth $60/year in Patreon subscriptions is up to you. But the option exists, and the shows that take it consistently deliver the lightest ad experience in our sample.

What we got wrong, or might have

Whisper's tiny model is fast but rough. Transcription quality suffered on conversational multi-speaker audio, episodes with background music, and anything not in English. Some dynamically inserted ads probably fell into untranscribed gaps and went undetected. Our numbers likely undercount ad time slightly as a result.

Our sample is 128 episodes from PodRanker.com's curated list of 500+ RSS feeds. It's not random in any statistical sense. It probably skews toward established, active podcasts and underrepresents both tiny indie shows and the massive commercial productions from Spotify and Amazon. The true distribution across all two million active podcasts could look different.

Borderline calls exist everywhere. When a guest mentions their company during an interview, is that a promotional segment or just conversation? When a host spends 8 seconds saying "brought to you by Closet by Design of St. Louis," is that really the same category as a 3-minute scripted BetterHelp read? We classified everything on a spectrum, but the lines we drew were judgment calls. Someone else would draw them differently.

And because of dynamic ad insertion, our numbers reflect one specific download from one specific location at one specific time. They're not universal truths about each show. They're measurements of a moment.

How podcasts compare to other media

Our numbers only cover podcasts, but it's worth putting them in context. Here's how the median podcast ad load stacks up against other media:

Medium	Typical ad load	Ads per hour
Podcasts (this study, median)	3.7%	~2 min
Spotify Free (music)	5–8%	~3–5 min
Streaming TV (ad tiers: Hulu, Peacock, Netflix)	6–8%	~4–5 min
YouTube (ad-supported)	10–15%	~6–9 min
Cable TV	25–30%	~15–18 min
Broadcast TV	27–33%	~16–20 min
Commercial radio	25–33%	~15–20 min

The gap is striking. A typical podcast has roughly one-eighth the ad load of broadcast TV or commercial radio. Even the ad-supported streaming tiers from Netflix and Hulu, which market themselves as a lighter alternative to cable, run about twice the ads of a median podcast.

YouTube is harder to pin down because it varies so much by video length and creator. A short video might have a single unskippable pre-roll that represents a high percentage of the runtime. A longer video with mid-rolls can hit 15% or more. But the average YouTube experience is still roughly 3-4x heavier than the average podcast.

The one thing these numbers don't capture is skippability. YouTube gives you a skip button after 5 seconds. DVR lets you fast-forward through TV ads. Streaming services are experimenting with interactive ad formats you can dismiss early. Podcast ads, by contrast, are baked into the audio. You can scrub forward manually, but there's no universal skip button, and most podcast apps don't offer one. Third-party tools with SponsorBlock-style features exist, but they're niche.

This matters because advertisers know it. Podcast ad completion rates are significantly higher than other media, which is why podcast CPMs (cost per thousand impressions) are among the highest in digital advertising. Listeners hear fewer ads, but they hear almost all of them. The lower volume is partly offset by higher attention per ad. Whether that's a fair trade depends on your perspective. As a listener, you're getting the lightest ad load in media. As a product of the advertising, you're one of the most captive audiences.

The worst podcast episodes in our sample, the ones hitting 24-27%, land roughly where normal cable TV lives. That's the ceiling for podcasts and the floor for traditional television. Even the heaviest podcast ad loads would feel unremarkable on cable.

What this tells us about podcasting

Most podcasts are pretty reasonable about ads. The median episode is under 4%, which beats commercial radio and ad-supported streaming video by a wide margin. The medium still offers one of the best content-to-advertising ratios in media.

But the long tail is real. If you listen to shows on major networks, or shows that lean into native advertising formats, you could easily spend 15-25% of your listening time on promotional content. Over a year of daily listening, that's weeks of ads.

What struck us most was the sheer variety. Podcasting has no single advertising model. It has dozens. Pre-roll, mid-roll, host-read, dynamically inserted, native, cross-promotional, Patreon, donation-based, and none at all. Every show picks its own combination, and the resulting listener experience ranges from completely clean to aggressively commercialized. No other medium gives creators this much freedom to choose, and the data shows they're choosing very differently from each other.

If ad load matters to you, the pattern is clear: smaller independent shows and listener-supported podcasts are consistently the cleanest. They might ask you for a few dollars a month, but the data says you get something real in return.

The data

All 128 reports are available as structured JSON: timestamped ad blocks, advertiser names, ad type classification, percentage breakdowns. If you want to dig into it or build on it, it's there.