What Does the Log Show?
On October 15, 2026 Podnews published an image of a request log from OP3, a prefix-based podcast analytics service, for the file podnews261006.mp3. It lists three GET requests at 01:36:42.099Z, 01:36:42.894Z and 01:36:43.138Z, all from the same hashed IP address and an Australian edge location. The first uses user agent "Podcasts/4027.210.1 CFNetwork/3896.200.61 Darwin/27.2.0" and range bytes=0-1. The second uses "AppleCoreMedia/1.0.0.24B5110a (iPhone; U; CPU OS 27_2 like Mac OS X; en_au)" with the same range. The third uses the AppleCoreMedia user agent and range bytes=1277952-3374582 (Podnews log image).
bytes requested (computed from the Range header)
Podnews's log gives the Range header values, not byte counts. We computed the third value as 3374582 minus 1277952, plus 1, because HTTP byte ranges are inclusive. The "bytes served" field is not in the image. This is one log of one device and does not show that every Apple Podcasts play behaves this way.
The chart shows that two of the three requests ask for 2 bytes each while the third asks for roughly 2.1 million bytes, so only one request retrieves a meaningful slice of the episode.
The log is one device on one day. It does not show how often the pattern occurs, how Apple Podcasts behaves on other devices or versions, or what any host's analytics reported for the same play. Podnews's October 15 item says that when an episode is played in Apple Podcasts without being downloaded first, the app "appears to make three download requests," and that it saw the behavior on the iOS 27.2 beta and in the public iOS 27 release (Podnews). That is a publisher's test, not an Apple statement. Apple has not, in anything we read, described this behavior.
What Does IAB Tech Lab's v2.3 Say About Such Requests?
The finalized guideline addresses two-byte requests directly. In section 5.5.2, step 1.2, item 6, it says: "Some apps perform a 2 byte range request (Range: 0-1)" and "Disregard any 2 byte (0-1 byte) range requests. These requests don't represent a valid download." Section 7 asks apps to request a progressive download "in slices at a byte range that is more than 2 bytes at a time." On the log above, the guideline's rule would leave the third request as the only candidate (IAB Tech Lab, v2.3).
Whether the third request becomes a counted download depends on the next step. Section 5.5.3 sets the threshold at "enough of the podcast content to play for 1 minute," and allows the "bytes served" value to be used to determine how much was downloaded. The log image does not include bytes served, so we cannot say whether this request met the threshold. IAB Tech Lab finalized v2.3 on September 29, 2026 after a public comment period from July 21 to August 19, 2026 (PPC Land); its own page shows "Last Updated: September 29, 2026" ([IAB Tech Lab](https://iabtechlab.com/standards/advanced-tv/podcast-technical-measurement-guidelines-v2-3/)).
Why Does the AppleCoreMedia Label Matter?
Podnews says that under v2.3 the only request that counts is "the one marked" AppleCoreMedia, and that Apple Podcasts has not used that user agent "since Mar 2023" (Podnews). The first half follows directly from the log: once the two bytes=0-1 requests are set aside, the third request, sent with the AppleCoreMedia user agent, is the only one left. The guideline does not name AppleCoreMedia or any other app; it asks apps to send "enough details in the user-agent header to allow it to be consistently differentiated from the user agent of other devices" (section 7, item 10) and to submit their user-agent values to the IAB Tech Lab Spiders and Bots inclusion list (section 7.1).
The second half is the practical problem Podnews raises under the headline "Is Apple Podcasts appearing correctly in your podcast statistics?" Hosts identify listening apps from the user agent. If the request that survives filtering carries a generic AppleCoreMedia label rather than the "Podcasts" label, a host's dashboard may not credit that download to Apple Podcasts. That is our reading of Podnews's point; Podnews does not say how any named host handles it, and Apple has not commented in anything we read. A related point, also our reading: v2.3 identifies unique downloads by IP address plus user agent (section 5.5.4), so a provider that failed to drop the 2-byte requests could count the "Podcasts" request and the AppleCoreMedia requests as two downloads from one play.
Why Does a Miscount Matter for a Placement?
A counting error that doubled or tripled requests would inflate a show's download numbers and any sponsorship rate based on them. The guideline already filters out the two-byte requests and states that its figures count downloads, not listens. Pipkin put it this way: "A download is not the same thing as a confirmed play" (PPC Land). See what changed in the podcast download rules and why podcast download numbers mislead.
What This Means for Anyone Buying a Podcast Placement
Download numbers are a seller's claim until the counting method is shown. Before buying, ask the host or network three things, each our recommendation rather than an IAB rule. First, does the analytics provider apply the v2.3 two-byte filter? Second, is the provider IAB Tech Lab certified, and which measurement window does it use? Third, is the figure downloads, unique consumers or platform plays? Ask for a platform-side number as well: Apple Podcasts and Spotify analytics are the platforms' own figures, which we compare with host numbers; see our guide for what each reports.
A paid podcast spot is advertising and should be disclosed as such. Nothing here suggests that a higher download count buys a higher audience.
Methodology and Limitations
Four limits apply. First, the evidence of Apple's behavior is a single log image from Podnews, which we read as an image; Podnews's own claims come from its October 15 newsletter item. Second, the byte count is our arithmetic from the Range header. Third, the guideline text we read comes from IAB Tech Lab's GitHub copy of v2.3, which carries no release date in the section we read, so the September 29 date comes from PPC Land and IAB Tech Lab's page. Fourth, we ran no test of our own and did not contact Apple, Podnews or OP3.
Conclusion
If the log is typical, the filter in the new guideline does what it says: two of three requests drop out. The uncomfortable part for a buyer is that this depends on every provider applying the filter correctly, and a show owner cannot always say which provider produced a quoted number. Ask for the method before the rate card.
Frequently Asked Questions
Does Apple Podcasts make three download requests for one play?
In a log Podnews published on October 15, 2026, one iPhone play produced three requests within about one second. That is one device on one day, and Apple has not described the behavior in anything we read.
Which request counts as a podcast download?
IAB Tech Lab's v2.3 guideline says to disregard 2-byte range requests, which it says "don't represent a valid download." A download is then enough content to play for one minute, or the full file for episodes under a minute.
What is a range request?
A range request asks a server for part of a file, identified by byte positions in the Range header. In the Podnews log, two requests asked for bytes=0-1, which is 2 bytes, and one asked for bytes=1277952-3374582.
Why does the AppleCoreMedia user agent matter?
Podnews says the AppleCoreMedia request is the only one that counts under v2.3, which matches the log once the 2-byte requests are dropped. Podnews also says Apple Podcasts has not used that user agent since March 2023, so a host may not credit the download to Apple Podcasts. The guideline itself names no app's user agent.
When did IAB Tech Lab finalize v2.3?
PPC Land reports September 29, 2026, after a public comment period from July 21 to August 19, 2026. IAB Tech Lab's page shows "Last Updated: September 29, 2026."
Does a download mean someone listened?
No. IAB Tech Lab's Brad Pipkin said, as reported by PPC Land, "A download is not the same thing as a confirmed play."
Sources
This guide is built from a Podnews log image, IAB Tech Lab's v2.3 guideline text on GitHub, IAB Tech Lab's standards page and a trade report on the finalization, all read on October 15, 2026. The buyer questions are our recommendation, not an IAB rule. We read Podnews's October 15 newsletter item for its own test and claims, and did not test any host's analytics.
- trade press Podnews, apple-podcasts-useragent-261015. Supports the three requests, user agents, ranges and timestamps. Read October 15, 2026.
- standards IAB Tech Lab, Podcast Technical Measurement Guidelines v2.3. Supports the 2-byte rule, one-minute threshold and user agent requirements. Read October 15, 2026.
- standards IAB Tech Lab, Podcast Technical Measurement Guidelines v2.3. Supports last-updated date. Read October 15, 2026.
- trade press PPC Land, IAB Tech Lab finalises v2.3 podcast counting rules covering video. Supports finalization date, comment period and Pipkin quote. Read October 15, 2026.
- trade press Podnews, Podcast hosts are your friends: but better. Supports Podnews's Apple Podcasts test (iOS 27.2 beta and iOS 27), the three requests and the AppleCoreMedia and March 2023 statements. Read October 15, 2026.