Do 90% of podcasts really quit after three episodes?

The podfade numbers everyone repeats have no primary source. This is a census of every living feed’s episode count — all 4,711,475 of them — so the folklore can be checked instead of repeated.

Census Vintage2026-08-17
live
Recomputed with the monthly Report cycle
Language
Episodes currently in each living feed — the 4,711,475 feeds this scope covers, with the dead list’s bare IDs necessarily excluded.

Episodes published

bar area = share of all feeds
101+
0102050100
0
226,9444.8%
1
977,65320.8%
2
326,0136.9%
3
300,7316.4%
4-7
814,65117.3%
8-10
370,5827.9%
11-20
580,31112.3%
21-50
526,37011.2%
51-100
280,6196.0%
101+
307,6016.5%
The survival ladder
61.1%made it past episode 3
43.8%past episode 7
23.7%past episode 20
Of the 4,711,475 living podcast feeds, 38.9% have published three episodes or fewer, and 23.7% have published more than twenty (PodHood Insights, Vintage 2026-08-17).

The folklore, checked

Three claims that circulate without a primary source, against the census.
“90% of podcasts quit after three episodes.”

38.9% of living feeds stopped at three or fewer. Even if every one of the 3,245,744 dead feeds had quit that early, the share of everything ever listed would top out at 63.8% — the 90% figure cannot describe this data.

“Most shows die at the seven-episode wall.”

Half true: 56.2% of living feeds have seven episodes or fewer. The wall is real, but it is a slope — no cliff appears at episode 7 in the distribution above.

“Reaching episode 21 puts you in the top 1%.”

23.7% of living feeds — nearly one in four — have published more than twenty episodes. The real top 1% sits far beyond a hundred episodes.

What this census can and cannot say

Episodes currently in the feed

The count is what each feed carries this Vintage. A show that trims its back catalogue — daily news feeds often keep a rolling window — counts low, so long-running shows are undercounted, never overcounted.

Living feeds only, and which way that cuts

The 3,245,744 feeds the index has marked dead are bare IDs — no episode counts. Dead feeds skew short, so the true quit-early share of everything ever listed is higher than the living-feeds figure — which is why the worst-case bound above is stated alongside it.

Where the folklore may come from

Survey samples and coaching-industry rules of thumb, compounded by repetition. None of the circulating figures cites a dataset; this page cites its methodology.

Cite this figure

PodHood Insights, "Podfade: how many episodes podcasts really publish", Vintage 2026-08-17. Retrieved from podhood.com/insights/podfade
The CSV carries the same Vintage stamp in its header, so a number lifted out of this page never loses the date it was true on.

More from Insights

All Insights →