The podcast industry is running the same experiment commercial radio ran in the 1990s: how much ad load can you add before the audience leaves? Radio found out. Audio's current generation is about to.
What we measured
Four hundred shows, eighteen months of episode-level data, ad loads extracted from the audio itself. Median ad load rose from 3.1 to 4.4 minutes per hour of content. And listener behavior bent at a remarkably consistent point: shows crossing 4.5 minutes per hour saw 90-day listener retention fall an average of 11 points versus their own pre-increase baseline. Below that line, retention was essentially flat regardless of load.
The composition effect
Load isn't the whole story — composition matters as much. Host-read spots showed near-zero skip correlation at any volume we observed. Dynamically inserted announcer spots drove 3× the skip-ahead behavior per minute. Shows that grew load purely through DAI hit the retention cliff earlier, around 3.8 minutes.
The uncomfortable math
A show at the cliff edge that adds one more mid-roll makes roughly 12% more revenue per episode this quarter — and, on our numbers, ends the year with a smaller audience than it started. The networks pushing load hardest are monetizing a decaying asset and reporting it as growth. The shows treating ad load as a retention variable, not a revenue lever, are the ones that will still have audiences to sell in 2028.