Video or Audio: How Are People Consuming Podcasts Now?

How many people watch rather than listen?

Edison Research reports that 57% of Americans aged 12 and over have both listened to and watched podcasts. Set against the 80% who have ever consumed a podcast in any form, that means roughly seven in ten of everyone who has ever touched the medium has encountered it as video at some point.

The figure is a lifetime measure rather than a habitual one, so it does not establish that most consumption is now visual. It establishes something weaker and still significant: that watching a podcast is a normal behaviour rather than a niche one, and that a format defined by its delivery mechanism has an audience that does not treat the mechanism as fixed.

The supply side has moved further than the audience. About 71% of US podcast creators produce a video version, per the note accompanying the IAB and PwC revenue figures. Production has shifted faster than consumption, which is the usual order: the video exists before the habit of watching it does.

The gap between the two audience figures is where the behaviour actually sits. Eighty percent have consumed a podcast in some form; 57% have done both. The 23-point difference is the population that has only ever listened — people for whom the medium remains what it was a decade ago, consumed in cars, kitchens and gyms where a screen is unavailable or unwanted. That group has not shrunk because video arrived; it is simply no longer the whole of the audience.

This is why framing the shift as a migration misdescribes it. Nothing in the data shows listeners converting into watchers. It shows a second mode of consumption arriving alongside the first, used by a majority, and leaving the original mode intact. The medium has added a surface rather than changed one.

Why does the advertising data still call it audio?

Because the category was defined before the behaviour changed, and revenue reporting follows the definitions sellers use in their own books. The IAB and PwC state plainly that their podcast figures reflect the traditional, audio-based definition of the medium.

That is a defensible choice for continuity — changing a definition breaks the year-on-year comparison that makes a revenue series useful — and it has a predictable consequence. Advertising sold against a video podcast on a video platform is likely to be counted as video advertising, because that is the seller's product and that is the line it is booked under.

The result is a category reporting $2.862 billion in 2025 US revenue while a majority of its creators produce output that may be counted elsewhere. Neither body publishes an estimate of the gap, which is why this report does not offer one.

There is a practical test of how much this distorts the picture. If video podcast inventory were being counted as podcast revenue, the category's growth rate would be expected to track the growth of video advertising, which ran at 25.4% in 2025. It ran at 17.6% instead, between video's rate and the 10.2% recorded for digital audio as a whole. That is consistent with a category whose fastest-growing segment is being booked somewhere else, though it does not establish it.

The definitional problem also affects who is credited with the audience. A show distributed on a video platform accrues subscribers and watch time to that platform's account rather than to a podcast feed, so the same programme can look substantial by one measure and modest by another depending on which system is consulted.

Does the distinction matter for a guest appearance?

It changes what an appearance produces. An audio-only recording yields an episode and a transcript. A video recording yields those plus a filmed asset that can be cut into short clips, captioned, and posted to platforms whose distribution does not depend on the podcast's own audience at all.

That difference is the reason the size of a show's listenership is a weaker predictor of an appearance's reach than it used to be. A modest programme that clips well can put a guest in front of more people through derivative assets than a larger audio-only show reaches through its feed, and the clips persist and recirculate long after the episode's release week.

It also changes the preparation. An appearance recorded on camera is judged on how it looks as well as how it sounds, and the thirty-second excerpt that travels is selected for its self-contained clarity rather than for its place in the conversation. Those are different performance demands from a radio interview.

The economics of the two formats differ for the host as well as the guest. Video production carries costs audio does not — lighting, framing, editing, an editor who works in a visual timeline — which is one reason the creator figure of roughly 71% is striking. Producers are absorbing a materially higher cost base because the distribution advantage of clips is large enough to justify it.

For a guest, that cost asymmetry is worth knowing at the point of booking. A show investing in video has a reason to want the appearance to look good, and will generally ask more of a guest's setup than an audio show does. Appearing on camera from a poorly lit room on a laptop microphone produces an asset the host is less likely to promote, which quietly removes the distribution advantage the appearance was sought for.

What does this mean for measurement?

It means most published figures about podcasting should be read with their definition attached. Audience figures from Edison count people who consume podcasts by any means, including watching. Revenue figures from the IAB and PwC count money booked against an audio definition. Those two series describe overlapping but different objects, and dividing one by the other — revenue per listener, for instance — produces a number with no coherent meaning.

It also means claims about the medium's growth need a stated denominator. The audio category can look like it is growing slowly while the underlying activity grows quickly, simply because a rising share of the activity is being recorded under a different heading.

The honest summary is that podcasting's audience is measured reasonably well and its economics are measured by a definition its participants have outgrown. Until the revenue series is redefined, or until someone publishes an estimate of what the audio-only definition misses, the size of the medium is known within a margin nobody has quantified.

There is a narrower measurement consequence that matters when judging a specific show rather than the medium. Download counts, the currency in which audio podcasts are still sold, do not capture video consumption at all. A programme with modest download figures and a substantial video following is undersold by its own metrics, and one quoting combined reach across both is producing a number no standard defines.

None of this is an argument against appearing on podcasts, and it is not an argument for preferring video shows. It is an argument for asking a specific question before agreeing to anything: where will this be published, in what formats, and which of those does the show actually promote. The answer determines what the appearance is worth far more reliably than any figure about the medium as a whole.

The medium is, in short, better understood than it is counted, and the counting is improving more slowly than the behaviour is changing.

Methodology and limitations

Audience figures are from The Infinite Dial 2026, produced by Edison Research at SSRS, fielded January 2026 among 2,050 Americans aged 12 and over by random-digit dialling and weighted to the national 12+ population, sponsored by SiriusXM Media. The creator figure and the definitional note are from the IAB and PwC Internet Advertising Revenue Report for full year 2025, published 16 April 2026. Four limits apply. First, the 57% figure is a lifetime measure of having both listened and watched, not a measure of how consumption is currently split, and it should not be read as one. Second, the audience and creator figures come from different studies with different populations and are charted together only for comparison, not as a single series. Third, the extent of revenue misclassification under the audio-only definition is not quantified by either body and is not estimated here. Fourth, no figures on the share of listening time spent watching are cited, because no source reviewed here publishes them.

Conclusion

A majority of Americans who consume podcasts have watched as well as listened, and roughly seven in ten US creators now produce video. The advertising figures that describe the medium's size continue to count it as audio, by the explicit statement of the bodies that publish them. Both facts are defensible in isolation and together they leave the category's economics measured by a definition its participants have moved past. For anyone appearing as a guest, the practical consequence is not about definitions at all: the recording will most likely exist as video, the clips will outlive the episode, and the appearance should be prepared for accordingly.

Frequently Asked Questions

Sources

1. vendor research Edison Research at SSRS, 2026. The Infinite Dial 2026, fielded January 2026 among 2,050 Americans aged 12+, sponsored by SiriusXM Media.
2. trade body Interactive Advertising Bureau and PwC Advisory Services, 2026. Internet Advertising Revenue Report, Full Year 2025, published 16 April 2026, including the note on the audio-based podcast definition and the share of creators producing video.
3. trade press Radio Ink, 2026. “IAB: Digital Audio Grew 10% In 2025 as Podcasts Near $3B”, 16 April 2026.