For decades, podcast advertising has been bought and sold against downloads—a proxy that measures file delivery rather than human attention. The industry’s complicated relationship with the download is well-documented, including in thesepages, but the conversation has evolved. This is no longer simply a measurement complaint. It’s a business opportunity.
That opportunity truly “amped up” last week. The AMP Accords, a ratified measurement standard from the twelve-member task force convened by Oxford Road’s Dan Granger and shaped in no small part by his colleague Giles Martin, was formally released and now is in the court of industry opinion. Dan and I have been comparing notes and wrestling with this necessary shift for a while, most recently on Oxford Road’s Media Roundtable podcast on June 26.
AMP is good, and I’m glad it’s here.
Podcasting is becoming an inherently multi-platform business — audio, video, clips, each measured differently — and a shared standard isn’t a nice-to-have anymore. It’s what keeps this category competitive against every other medium chasing the same ad dollars. Getting twelve companies from competing corners of this business to meet monthly for a year, vote anonymously, and finish with three ratified proposals is far harder than it looks. Most standards efforts die of politeness. This one didn’t — in large part because of the task force’s smartest procedural rule: a “no” vote came with homework. You couldn’t simply object; you had to arrive with something better. That rule should outlive the group.
Why the Download Had to Go
A download is a file delivered to a device, not proof that anyone listened; auto-downloads accumulate on phones without ever being played. Dan Misener and Jonas Woost, co-founders of Bumper (who spoke to my NYU class last semester), use a show-of-hands test at industry events: how many people have podcast episodes downloaded that they’ve never listened to? Almost every hand goes up. Bumper’s estimate is that selling on downloads may overstate actual audience by as much as 50 percent.
Whether the real gap is 50 percent or smaller, the structural point holds. A number that counts file delivery is antiquated and retarding growth of the business.
Apple has already lived a version of this reckoning. Its 2023 iOS 17 update tightened how the Podcasts app counted a download, and totals for many shows dropped overnight — enough to shock publishers who had built revenue projections on the old number. That wasn’t an attack on the download; it was Apple trying to restore confidence in the metric itself, filtering out exactly the kind of auto-download noise Bumper’s show-of-hands test surfaces. It made the download more honest, but it didn’t make it the right metric. AMP isn’t patching the download the way Apple did — it’s proposing to leave it behind for something else entirely.
A couple of key points on what’s actually in the document, briefly. It defines a podcast as an on-demand, spoken-word show intelligible with your eyes closed — video and RSS are both allowed, neither required. And it replaces the download with four exposure metrics — Play, Audience, Ad Impression, Ad Audience — meant to sit alongside the old count rather than replace it overnight; the implementation guide behind them, a real spec rather than a policy statement, was authored by Pete Birsinger of Podscribe.
Sometimes it helps to look outside an industry to get a fresh take.
Television Already Wrote the Playbook
I’ve spent real time with people who run television stations and cross-platform broadcast businesses, and the parallel to podcasting’s moment is striking. For fifty years, TV was sold on ratings, gross rating points, and average quarter-hour audience — metrics that worked when everyone watched the same content at the same time. Then audiences fragmented across streaming, OTT, and dozens of apps, and while the industry tried, it didn’t invent a better rating. Instead, it adopted a common currency: the impression, verified regardless of which screen it landed on.
Pierre Bouvard, chief insights officer for Cumulus Media | Westwood One, points out that every new medium starts the same way: cable launched selling “homes passed” whether anyone subscribed, early TV measured households instead of people, and Nielsen’s People Meters and later C3 kept narrowing the count toward actual ad exposure. Each correction shrank the reported audience and grew the revenue. Podcasting is approaching the same moment.
The formula is simple: trust the numbers, grow the revenue. It’s that simple.
Measurement Changed the Business
Television’s transition was about one currency — cost per thousand verified impressions — across every surface: linear, streaming, YouTube pre-roll, or an app. Hearst Television led the move in 2019; Nexstar, NBC’s “One Platform” and the Television Bureau of Advertising followed, backed by cross-platform dashboards showing clients impression pacing and delivery in real time.
The business impact was real. Stations were no longer boxed in by a single newscast’s spot inventory; they could guarantee impressions across every platform their content lived on. The revenue ceiling became total audience, not ad inventory. That is the argument I most want podcasting to absorb, and it is the one I hear least often in this conversation.
Essentially, that is the roadmap the Accords are pointing at — and it is the right next step for podcasting.
“The money is going to flow to where the audience is. We have the audience. We have great shows. If we can verify the audience and present it in a unified sort of way, that’s where we need to be.”
— Steve Goldstein, Oxford Road Media Roundtable
Three Forces Are Accelerating the Shift
Three structural changes make this urgent, not just philosophical.
Podcasting is becoming a video business, and video doesn’t download.
A single podcast now lives as audio, YouTube video, and clips, each measured differently: a YouTube view is server-verified and consumption-based, a download gets logged whether anyone plays it or not. That gap grows every month as video’s share of listening increases.
Media buyers increasingly see podcasts as two different products.
An agency meeting I helped organize in New York surfaced the issue plainly: audio and video buyers are often different teams with separate budgets. Podcasts spanning both formats risk falling between those silos unless presented as one audience product. How that gets resolved will shape how podcast ad revenue scales — and it is precisely the problem the Accords’ definition was written to solve.
The money is sitting in the dashboard.
The largest ad budgets are planned inside platforms showing impression delivery, pacing, and real-time verification — television, digital video, streaming audio. Podcasting mostly isn’t in those systems, so it sits outside the industry’s largest media allocations. Getting onto those dashboards, in the same impression-based language as CTV and streaming audio, is estimated to unlock $1–2 billion in additional ad revenue. The audience is already there. The money is waiting for the infrastructure.
What’s a podcast …..one more time …
The line most likely to be remembered is the shorthand for defining a podcast: if it works with your eyes closed, it’s a podcast. The group had the discipline to boil it down to a bumper sticker.
That may be right for an industry definition, but as always, the audience is the arbiter of what a podcast is, and always has been. In our Coleman/Amplifi study last April, 85 percent said a podcast can be audio or video — no test, no debate — and every survey since has pointed the same way. That isn’t a flaw in the definition. It was a change driven by consumers.
“Podcast” has become the culture’s shorthand for a spoken-word conversation show — a few people talking, at length, in a way you trust — wherever it happens to live. Podcasting is having a genuine cultural moment, and it’s a proxy for that kind of conversation now, regardless of format.
So there are two definitions, and both are right. The consumer’s is broad, and it’s the one powering the cultural moment. The industry’s is narrow, and it exists for a reason Dan Granger stated plainly on the Podnews Weekly Review this week: the consumer can call a podcast whatever they like, but the people buying and selling it have been stuck. A channel nobody can classify is a channel no department owns — is it audio’s, social’s, video’s? Some agencies still file podcasting under their offline division. Money with no owner doesn’t get spent, and that has quietly cost this business real revenue.
Thirty Seconds Is a Starting Line, Not a Finish Line
A play is thirty consecutive seconds, audio or video. That is a real standard and it is the right place to begin. Spotify has already adopted it, YouTube is understood to be in the same neighborhood, and this means two of the largest players in podcasting are roughly speaking the same number for the first time. That is not nothing. Now Apple needs to get on board, and Apple has never shared data of this kind. The various other apps carrying roughly 30% of RSS listening need to click yes.
Some apps say they want a cut of revenue. Maybe the ones doing real engineering to report compliant data should. But look at web browsers — Chrome and Safari are free, do real engineering, and carry an ocean of ad-supported content, and no website pays them a cut for rendering the page. (Google does pay to be the default search engine in some of them — a fee for a specific placement, not a tax on every site’s ad revenue.) Distribution alone isn’t a claim on the ad dollar.
Back to the issue of thirty seconds. Is it enough? Not quite. But it’s a great and essential starting point.
The Questions That Will Decide How Much of This Turns Into Money
Directionally, the Accords are the impression model, even though the document never uses the word. Separating content exposure from ad exposure is the same architecture that let television get bought on the same terms as the rest of a digital media plan. The roadmap is now written down. What happens next decides whether anyone drives it.
Does the buy-side infrastructure get built?
Television’s shift paid because it arrived with cross-platform dashboards showing advertisers pacing and delivery in real time. Standardized numbers that live in PDFs and one-off reports do not move money. This is the harder, less glamorous half of the job, and the Accords are largely silent on it.
Can podcasting get onto those screens without being sold like the rest of digital?
This is the one I feel most strongly about, and it’s where I’d push back on anyone reading my television comparisons as an argument for treating podcasts as generic inventory. TV won the impression fight, and a great deal of TV inventory became interchangeable in the process. Much of podcast advertising will work well in this environment. But not all.
What about the host-read ad? Does it get crushed?
No. Podcasting can’t afford to become purely transactional, and the host read is exactly where that risk is lowest. What earns this medium its premium is host trust and intimacy, not delivery efficiency — and advertisers keep buying the embedded host read because it’s unique and effective in ways a served ad can’t replicate. That value should be priced on its own terms, not folded into a generic impression rate.
The Accords don’t say everything has to be priced as an impression — and they don’t protect against it either. Worth flagging now rather than after the fact. None of this is a knock on the work. The Accords don’t have to resolve everything today. A common language for the industry is job one.
On Pacing: Shorter Runway, Firm Date
There’s a fourth place I’d push, and it’s about pacing. The framework runs plays alongside downloads for close to a year (or more?) before pricing moves. Run that interim too long and the risk isn’t caution, it’s attrition. With no deadline forcing the question, the players who haven’t integrated yet have every reason to wait it out — and the longer shadow reporting runs, the more momentum it burns.
I lived through a version of this as Chairman of the Arbitron Advisory Council (later purchased by Nielsen) as radio moved from diary to PPM electronic measurement. The public comparative testing that actually mattered took about two months, not a year, and moving quickly once the case was made and the plumbing perfected is what built market confidence. A long parallel period doesn’t make a currency safer. It delays the moment buyers trust it. The IAB Tech Lab opened its own podcast measurement guidelines for public comment two days before the Accords landed, still built on server-side download counting. Two standards moving in opposite directions in the same week is exactly the fragmentation this work was meant to end.
Shorter runway. Firm date.
The Next Twenty Years Require a Different Currency
The advertising business is increasingly operating in one common language across every medium, and verified impressions are how agencies compare audiences and allocate budgets across platforms.
The download gave podcasting twenty years of remarkable growth, but it isn’t the right tool for the next twenty. Television spent nearly a decade learning that lesson; streaming and digital video already have. Publishers who moved first, building the infrastructure to verify delivery, captured the upside — more addressable inventory, more revenue. Better data, more dollars. That’s the whole case.
Podcasting doesn’t need to invent a roadmap. The AMP task force just handed it one. The open question is whether this industry has the patience to build the plumbing and the nerve to move faster than the timeline says — because unlike television, podcasting does not have a decade.
