Briefing: The Industry Finally Agrees on What a Podcast Is

Last week podcasting booked a room next to the Upfronts. This week it showed up with a rulebook. Two measurement efforts landed inside a few days of each other, and between them they take a swing at the oldest problem in this business: nobody could fully agree on what a podcast is or how to count one. Buyers have been pricing an inventory they could not measure the same way twice. That is the thing now on the table.

The AMP Accords put a definition on the table

The Alliance for Measurement in Podcasting, a group of twelve companies convened by the agency Oxford Road, published what it calls the first cross-platform measurement standard for the medium. The document was walked through at Oxford Road's CAO Summit on July 23 and covered this week by Podnews. The membership is not a fringe committee. It includes Spotify, SiriusXM Media, Libsyn, Shopify, DraftKings, BetterHelp, United Talent Agency and Podscribe, which is to say platforms, advertisers, a talent agency and a measurement vendor sitting at one table.

Three things in it matter for money. First, a definition: the Accords propose that if a piece of content works with your eyes closed, it is a podcast, audio or video, RSS or not. That sounds like a slogan, and Radio Ink ran it as one, but it is doing real work: it folds video podcasts into the same measurable category instead of treating them as a separate YouTube problem. Second, a new exposure metric they call a Play, which only counts once someone has actually listened to or watched at least thirty seconds. That is a deliberate shot at the download, a number that has always included files that were fetched and never heard. Third, an attribution approach built on randomized holdout groups, the same incrementality method the big walled gardens already use. The pitch attached to all this is roughly a billion dollars in ad spend the group argues is sitting on the sidelines waiting for numbers it can trust.

One agency's public response, and where it pushes back

Worth reading alongside the framework: the podcast advertising agency Adopter Media published a detailed response after an early look at the document. The tone is supportive, calling the Accords an excellent first step, but it lists nine things it says are still missing, and two of them are the ones a buyer should think hardest about. One is that the closed platforms are further from compliance than they look: YouTube and Apple do not currently expose the event-level consumption logs the AMP specs assume, so a standard built on that data has a hole exactly where the biggest audiences are. The other is privacy. Riding advertising IDs and hashed emails alongside listening logs, the response argues, is a real escalation in how identifiable podcast listening becomes, and it makes the case for baking privacy terms into the standard itself rather than leaving them to side contracts. Neither point sinks the Accords. Both are the kind of thing that decides whether a framework gets adopted or just admired. We link it because it is one of the more specific reactions on the record, not as an endorsement; the desk has no stake in AMP or in any agency's read of it.

The IAB moved the same week

Not to be lost in the noise: IAB Tech Lab released version 2.3 of its Podcast Technical Measurement Guidelines for public comment, its first revision since February 2024. The headline change is a vocabulary one that carries more weight than it looks: "listener" becomes "podcast consumer," a nod to the fact that a meaningful share of the audience is now watching. It also adds guidance on URL-prefix measurement, RSS enclosure changes and invalid traffic, and the Lab has said it is groundwork for a v3.0 in 2027 aimed squarely at streaming video podcasts. It is open for comment for thirty days, until August 19. Two standards bodies pointing at the same target in the same week is not a coincidence. It is the industry deciding that the measurement mess is the thing blocking the next leg of ad growth.

Meanwhile the money did something interesting

While everyone argued about counting, the spend held steady and got more efficient. Magellan AI's numbers, recapped this week, put the top fifteen advertisers at about $361 million across the first half, with June's top fifteen at $60.3 million, a hair under May. Quince led again at $6.9 million, its third month at number one this year, ahead of Shopify at $6 million, Toyota at $5.7 million, BetterHelp at $5.5 million and T-Mobile at $4.1 million. The same week Acast reported its Q2, and the shape of it is the whole argument for better measurement in one line: revenue per listen climbed 26 percent to a company record while total listens grew just 2 percent, as ppc.land noted. Growth is coming from monetizing the audience better, not from a bigger audience. That only continues if buyers believe the numbers underneath the rate.

What the desk would do this week

Buyers: read the Play definition and do the arithmetic on your own flights. If a listen has to clear thirty seconds to count, some of the reach you are paying download CPMs for today gets reclassified as never-heard, and your effective cost per real exposure goes up on paper even though nothing about the show changed. Better to model that now than to be surprised when a vendor adopts it. And send a comment to the IAB before August 19; this is the rare window where the people spending the money get a vote on the ruler. Sellers: get your consumption data in order, especially channel-level numbers from YouTube and Apple, because a standard that rewards shows people actually finish is a standard that rewards you if you can prove it. The download hid a lot of sins. The Play will not.