Does This Podcast Interview Guests?

What a show says, against what its titles show

Every readable feed placed in exactly one of four cells, then run through the other two checks we apply everywhere else on this site.

The top row is the whole argument. 11 shows out of 362 both say they book guests and prove it in a recent title. They are, unsurprisingly, the best targets in the corpus on every other measure: 100% still publishing, 100% publishing a contact route, a mean openness of 80 against a corpus median of 48. If that were the list, it would be a very good list and it would also be 11 shows long.

The two middle rows are where the real targets are, and they are the rows a single check throws away. 59 shows say they interview guests and their recent titles do not show it. 43 credit guests in their titles and never mention the format in the blurb.

Neither check finds most of the list

Stated as coverage: read only the blurbs and you find 70 of the 113 shows with any visible evidence, which is 62%. Read only the titles and you find 54, which is 48%. Run both and you find all 113, because that is what the total is made of.

The conditional version is blunter. Of the 70 shows that tell you they interview guests, 16% credit one in a recent title. Of the 54 shows crediting guests in recent titles, 20% say so in the description. Whichever direction you read it, the second check mostly disagrees with the first.

The reason is that the two artefacts are written for different purposes and at different times. A description is written once, near launch, and it sells the subject: what the show is about, who it is for. Format is an implementation detail the host already knows, so it often never gets typed. An episode title is written weekly, under a deadline, and it sells that episode: sometimes that means the guest's name, and sometimes it means the hook, with the guest credited in the show notes where a title check will never see them.

So the disagreement is not evidence that shows are lying about themselves. It is evidence that neither artefact was ever meant to answer this question, which is exactly why doing one of them and stopping is expensive.

The check that looks strongest ages worst

One column in that table runs against expectation, and it is the one worth changing your process over. Titles that name guests feel like harder evidence than a sentence in a blurb, because a name is a booking that happened. But the shows in that cell are the LEAST likely of the four to still be publishing: 60% against 77% for the readable corpus as a whole, a gap of 17 points.

That is not a coincidence and it has a mechanism. A feed always shows you its newest episodes, so a show that stopped two years ago still presents five titles with five guests in them. The evidence does not decay. The show does. And a description that says "we interview founders" is a statement about a plan, which is worth less on the day it is written and does not rot the same way.

The corpus-wide version of this trap has its own page: shows that publish visible guest evidence are not less likely to have stopped than shows that do not (27% of the evidenced ones had gone quiet, against 21% of the rest). Looking bookable and being live are unrelated properties, and research surfaces the bookable-looking ones first.

The practical consequence: whichever of the two checks you run, run the date check after it. Across the whole readable corpus that one removes 84 shows, more than any other check we apply.

What the bottom row does not mean

249 shows, 69% of everything readable, gave us neither signal. It is the biggest cell in the table by a distance and it is the one most likely to be misread, so here is the honest reading of it: we could not tell. It is not 249 shows that refuse guests.

Two ordinary things land a show in that row. Its blurb describes the subject rather than the format, and its titles carry the topic rather than the name ("Why your pricing is broken" instead of "Pricing, with a named guest"). Both are normal editorial choices, and a show can make both while booking a guest every single week. The cell is a limit on what a stranger can verify from a feed, not a verdict.

It is also not a cell to pitch blind. 79% of it was still publishing and 83% publishes a contact route, which is to say it looks exactly like the rest of the corpus on everything except the question you care about. If a show in this row is a genuinely good fit for you, the answer is thirty seconds in its episode list or its show notes, not a guess.

And that is a claim we can now put a number on rather than assert, because we went and read the notes for the whole corpus. 52 of the 248 shows in this row that publish notes credit a guest by name in them.

We ran the show-notes check ourselves

The obvious next move on that row is to open an episode and read its notes, which is what everyone tells you to do and what this page told you to do until we could say how often it works. So we read them: the recent episode notes for 360 of the 362 readable feeds, 99% of them, at a deliberately strict bar. Notes have to credit a person BY NAME, next to a booking phrase, before we count them. The word "guests" on its own is not enough in six hundred characters of prose the way it is in a title, because prose says it for a dozen reasons.

The top row is the finding. 52 of the 248 shows that neither check could see credit a named guest in their own show notes, which is 21% of the cell this page spent a section explaining. That is not a rounding error on a research method: it is roughly one in 5 of the shows a reader would have written off, and the evidence was published by the show the whole time, one click deeper than anybody looks.

The second row is the control, and it is there because a high number in the first row is equally consistent with a check that fires on everything. Of the 112 shows the blurb or the titles already flagged, the notes agree on 47%, about twice the rate they fire on shows nothing else could see. A check that found the two groups equally often would be measuring something other than guest bookings.

Counted across the whole readable corpus, the three checks together find 165 shows against 113 for the two published ones: 46% of readable feeds instead of 31%, a gain of 15 points. We are deliberately NOT folding it into the headline figure on this site, because that figure is quoted on other pages and a definition that changes underneath a sentence is worse than a conservative one. Read it as what it is: the size of the thing two checks cannot see, measured rather than assumed.

How to actually check, in order

The whole process, cheapest first, and it is under a minute per show:

• Read the last episode date before anything else. It is two seconds and it removes more shows than every other check combined.
• Read the blurb for format language: interviews, guests, conversations, sits down with, joined by. That is the check that finds 62% of the evidenced shows on its own.
• Read the last five episode TITLES for a person who is not the host. That is the other 48%, and it overlaps the first far less than anyone expects.
• If both come back empty and the show is still a good fit, open one episode's show notes. We ran that check across the whole corpus: 21% of the shows neither other check could see credit a guest by name in their notes, which makes it the most productive thing you can do with the next thirty seconds.
• Only then decide. A show that passes either check, is still publishing, and publishes a contact route is a real target, and there are more of them than either check alone will show you.

Applied to our whole directory, that sequence leaves 76 shows out of 384. That is the number worth internalising: the list is smaller than the directory and much bigger than whatever one check hands you.

Sources and benchmark date

Every figure above is derived at build time from our own read of the public feeds behind 384 podcasts between July 16, 2026 and August 26, 2026, not typed in. The method, the limitations and the raw data are public:

• Which Podcasts Actually Book Guests? (our study)

  Original analysis of the 362 readable feeds inside a 384-show directory. Source for the openness score and for both halves of the guest-evidence definition used here. Method, limitations and the full dataset are published with it under CC BY 4.0.

• Is This Podcast Still Active?

  The liveness check in full, and the corpus-wide version of the ageing problem: 84 of 362 readable feeds had published nothing in six months, and the ones that look most bookable are not the ones most likely to be live.

• How to Contact a Podcast Host

  The next check after this one: which door each show actually publishes, measured across the same feeds, and which of them is worth trying first.

• How Many Podcasts Should You Pitch?

  The same checks applied in sequence to the whole corpus: 76 of 384 shows passed all of them, which is what should size your list.

Frequently asked questions