Podcast Guest Statistics

What is measured here

9 sections, each answered from the same corpus and each linked to the page that derives it in full. Every figure carries the denominator it came from and the day we read the feeds. The last one is the limitations, and it is the shortest way to know what these numbers cannot be used to say.

How many podcasts book guests

32% of the feeds we could read publish something a stranger can check that says they book guests: a named guest credited in a recent episode title, or a show description that claims a guest-led format. The two checks are independent, and they disagree more often than anyone expects.

The overlap is the finding. Only 14 feeds do both, which is 11% of every feed where either check fires. Run only the description check and you will miss most of the shows whose titles prove it; run only the title check and you will miss most of the ones that say so plainly. A guest doing this by eye almost always runs one of the two.

There is a third check, and it is the one that moves the number most. Opening the recent EPISODE NOTES of 382 of those feeds finds a named guest credited in 123 of them (32%), including 59 shows where neither of the checks above fired at all. Counting notes as evidence too takes the total from 124 shows to 183, 48% of the corpus. We keep it as a separate measure rather than folding it in, because nine published pages quote the narrower figure and a silent redefinition would rewrite all of them at once.

And the depth you scan decides what you find. Reading 40 episodes per feed instead of 5 finds 94 of the 204 shows that credit a guest anywhere in that window; the other 110 (53.9%) are invisible in a five-episode glance. The rate is not a property of podcasts. It is a property of how hard you looked.

How many guest seats exist, and who holds them

A show that books guests is not the same as a show with a seat for you. Across the 286 feeds that were still publishing when we read them, about 2,669 guest seats a year are visible in the episodes themselves. They are held by very few shows.

66 shows, 23% of the live corpus, account for every visible seat, and the busiest tenth (29 shows) holds 77% of them. Among the shows that have any seat at all the median is 31 a year. There is almost no middle: a podcast either has a guest chair built into its format and fills it constantly, or it does not have one.

The seats are scarce but they are not gated. 95% of every visible seat belongs to a show that publishes a contact route beyond a bare homepage, and shows WITH a seat publish one at 91% against 85% for shows without. Having a guest chair does not make a show harder to reach.

One split worth carrying: 24% of independent shows had a visible seat against 21% of network-run ones. The recognisable names are the ones least likely to have a chair.

How a podcast can actually be reached

Every guide on the internet lists the routes you COULD try. This is which one each show actually publishes, counted on the feed itself plus the contact page we could find from it.

An email address is the answer 75% of the time. Everything else is a rounding error, including the guest application forms that the booking-platform market is built around: 5 of 383 shows publish one. Only 46 shows (12%) publish no route at all.

Two splits that run against intuition. Shows publishing guest evidence are LESS likely to be doorless (15% against 18%), so "I could not find an address" usually also meant "this was not a good target". And network-run shows are EASIER to contact than independents (7% doorless against 19%) while being harder to book. The two properties point opposite ways and are constantly confused for each other.

How many are still publishing

25% of the feeds we read had published nothing for roughly six months on the day we read them. 19% had published nothing for over a year, which is the half where "between seasons" has stopped being a plausible explanation.

The trap is that dormancy is invisible to every other signal a guest researches on. A good back catalogue does not decay when a show stops. Neither does visible guest evidence: the dormancy rate among shows WITH guest evidence is 27%, against 24% for shows without, so the shows that look most bookable have gone quiet at least as often as the rest. 30 shows in this corpus are more than a year dead AND publish visible evidence they book guests. Every one of them looks like a great target and is not.

The good news is that the check is trivial and unambiguous. 59% of readable feeds had published within the last week, and among the shows that were not dormant the median gap since the last episode was 3 days. There is no judgement call here: read the date on the newest episode.

What survives all the checks at once

Each rate above is one property of the corpus. A guest building a pitch list meets them STACKED, and because the properties correlate you cannot multiply the rates together. So the conjunction has to be counted.

408 names in; 383 with a readable feed; 286 still publishing; 246 with a door; 81 out the far end. The two numbers worth carrying are 246 and 81: the first is the honest "worth an hour" list, the second is the confident one, and the gap between them is what a feed cannot tell you rather than what is unbookable.

Which shows book guests: size, pace and length

The standard advice is to start small because big shows are unreachable. The data says the opposite, and it says it on every measure at once.

Guest evidence barely moves up the ladder (31% to 36%, no real gradient). What moves is liveness: 32% of the smallest shows were still publishing against 97% of the biggest. So "small shows are easier" is survivorship, not access: the small shows in any list you build are disproportionately the ones that stopped.

Publishing pace runs the same way. Across 382 feeds with a measurable pace, guest evidence sits near the corpus rate in every band while liveness climbs steeply with frequency, so a weekly show is not a better guest bet than a monthly one for the reason people think. It is a better bet because it is more likely to still exist.

Episode length is the one axis that predicts guest evidence cleanly. Across 382 feeds whose median episode length we could measure (read October 10, 2026), the shortest band publishes guest evidence at 11% and the longest at 46%. The median episode across the corpus runs 34 minutes; the median among shows that pass all three checks runs 42.

Guest booking by subject

The spread between subjects is far wider than the spread between show sizes, which makes it the single most useful cut on this page if you are deciding where to spend a week of pitching.

Startups leads on guest evidence at 61% and Finance & Investing trails at 18%, a spread no other axis on this page comes close to. Read each subject's "Gone quiet" share next to its "Publish guest evidence" share: the subjects with the strongest guest signal are frequently the ones that have gone quietest, so a shortlist built on guest evidence alone lands you in the graveyard.

What a booked guest actually gets

Every page selling podcast guesting promises a backlink and an audience. We measured the first one directly, as a WITHIN-show difference, because every show links out and only the gap between its own guest and non-guest episodes carries information.

Across 14,165 episodes from 383 feeds (read October 10, 2026), 78% of guest episodes carry at least one outbound link against 71% of the rest, an excess of 0.72 links per guest episode. Real, and much smaller than the promise. The distribution is what matters: 109 shows link out more on guest episodes, 77 link out LESS, and 21 shows (5%) publish no outbound link on any episode at all. The median show's excess is 0.05 links.

So a link is not something guesting buys you. It is something a particular show does or does not do, and it is visible in that show's notes before you pitch.

What these numbers are not

Every figure on this page is a measurement of what a public RSS feed PUBLISHES, on the day we read it. That is a narrower claim than the one it is constantly rounded up to, and the narrower claim is the defensible one.

• "68% publish no visible evidence they book guests" is not "68% do not book guests". A show can interview an expert every week and credit them nowhere a feed reader can see.
• "12% publish no contact route" is not "cannot be reached". A host who answers direct messages and says so nowhere is counted in that row.
• "25% have gone quiet" is a claim about the feed on the read date, not a prediction that the show will never return.
• The corpus is 408 shows from one directory, not a random sample of all podcasts. It is skewed toward the interview-shaped subjects a guest would search, which is the population the page is for and also a reason not to quote it as an industry-wide rate.
• Two findings here are read from their own files on their own days (October 10, 2026 for the outbound links, October 10, 2026 for episode length), so they are dated separately from the feed window rather than folded into it.

Everything on this page recomputes from the committed dataset when the feeds are re-read. No figure above is typed into the page, which is the only way a reference like this stays true after the first refresh.

Sources and benchmark date

Every figure on this page is computed from our own reading of public podcast RSS feeds. The corpus was read between October 10, 2026 and October 10, 2026; the outbound-link and episode-length measures are separate reads with their own dates, stated where they appear. The underlying dataset is published under CC BY.

• Which podcasts actually book guests? (our study, with the dataset)

  The full report on the 408-show corpus, including the CSV and JSON downloads and the deposited dataset DOI.

• The podcast directory these feeds come from

  Every show in the corpus, browsable by subject, with its own feed facts on its page.

Frequently asked questions