Omny Studio offers analytics for Apple's Podcast app. But how?

Omny Studios can now track how long people have listened to a podcast for listeners using Apple’s Podcast app.

Anna Washenko, reporting in RAIN News:

Clients using the analytics will be able to track and report consumption data from streamed plays of podcast episodes listened to on the Apple Podcasts app. The data reflects where the audience tuned in, and where they skipped or stopped, within 30 seconds accuracy.

But how?

Here are some clues from the report:

1) “roughly 30% of all Apple Podcast plays can be tracked”
2) “…within 30 seconds accuracy.”

Here is how I think they can do it: they can only track 30% of plays because an episode needs to be streamed (rather than downloaded first for later listening) and that, I guess, is the percentage of plays from the Apple Podcast app that are streamed.
Why only streaming? When streaming the server sends small pieces (or chunks) of the whole episode to the client and I would guess that Omny Studio can track how much data of an episode has been sent from the server. The Apple Podcast app (probably using HLS) receives ~30 second chunks of audio data from the server and that is why they can only track skips and stops with a 30 second accuracy.

That’s my theory, anyone got any other ideas?