What a podcast does for a message
A recorded conversation carries meaning differently from a page of text, and the difference decides what belongs in one.
The medium is slow on purpose
A written statement is read in the order the reader chooses. A recorded conversation is heard in the order it was spoken, at the pace it was spoken, and almost always in one pass. That single difference governs everything else about preparing for one. A listener cannot glance back at the second paragraph while the fifth is still playing, so an argument that depends on holding three conditions in mind at once will not survive the trip.
The compensation is duration. A press notice is measured in hundreds of words; a single episode segment routinely runs longer than every written statement an organisation publishes in a month. That length allows a position to be explained rather than asserted, and it allows the awkward part of a position to be addressed in the same breath as the appealing part. Written communication tends to drop the awkward part because there is no room. Audio has room.
What travels well in audio
Three kinds of material travel well. The first is sequence: how something came about, in order, with the turns left in. The second is reasoning: why one option was taken and another was set aside. The third is limitation: what is not yet known, and what would have to change for the answer to change. All three are hard to compress into a written line and easy to say aloud.
What travels badly is any content that depends on being seen. Figures, comparisons between more than two things, and precise names all lose their shape in speech. A listener will retain the shape of a number long after losing the number itself, so a figure spoken aloud should be rounded to the shape you actually want retained, and it should be one figure rather than four.
The audience is not a crowd
A broadcast audience is a crowd; a podcast audience is a queue of individuals, each listening alone, usually while doing something else. Direct address works. Rhetorical crowd-pleasing does not, because there is no crowd to please and no reaction to feed on. The register that works is closer to explaining something to one attentive person than to addressing a room.
That register also sets the honesty threshold. In a room, a deflection is covered by momentum. In a pair of headphones, a deflection is audible as a change of tone, and the listener hears the change before they identify the reason for it. It is generally cheaper to answer a difficult question plainly than to spend the following ten minutes regaining the ground lost by not answering it.
