Podcast episode ideas

Most advice about finding podcast topics points you outward. Trending searches, what other shows in your category are covering, a keyword tool. That skips the largest supply of usable ideas you have, which is the hundred hours of conversation you already recorded.

Your back catalog is full of half-finished arguments, questions you deflected, claims that have since been settled, and subjects you keep mentioning without ever giving them an episode. Those ideas are better than anything a trend list can hand you, because you already know you find them interesting and your audience has already heard you start on them.

Here are eight ways to get them out. The first four run on your archive. The last four run on everything around it.

1. The threads you dropped

Every long conversation leaves loose ends. A guest says "that is a whole other episode" and you both laugh and move on. You start explaining something, a better question interrupts you, and you never come back. Those moments are pre-validated ideas. You were curious enough to raise the subject and self-aware enough to notice you were leaving it.

The hard part is retrieval. Nobody remembers minute 74 of episode 112. The cheap fix is a running note during your edit pass, where you write down every "we should come back to that" as you hear it. The faster fix is searching your transcripts for the phrases that mark a dropped thread: "another episode", "we will get to", "I will not go into", "that is a longer conversation", "remind me to".

Ten minutes of that across a hundred episodes will produce more ideas than an afternoon staring at an empty document. It is also the one part of this list a machine is genuinely good at. Tally indexes your whole episode archive and builds a page per topic showing every episode where that subject came up, each one deep-linked to the moment in the transcript. A page like that is a machine for finding dropped threads, because it puts the four times you almost explained something next to each other.

2. The follow-up episode

Hosts treat episodes as finished objects. Booked, recorded, published, done. But the clearest signal about what to make next is what happened after you published. Which episode got replies. Which one made a stranger email you a correction. Which guest said something that aged well, or badly, over six months.

A follow-up is not a rerun. It is the same subject with a year of new facts, and usually a different person in the chair. If you covered a lawsuit, cover the ruling. If you interviewed a founder at seed stage, interview them after the Series B or after the shutdown. The audience already carries the context, so you can start at the depth it took you forty minutes to reach last time.

3. Listener questions, especially repeated ones

Ask for questions directly and often, then pay attention to which ones arrive more than once. One person asking is a question. Four people asking the same thing in different words is an episode, and the title is usually the question itself.

You do not need a big audience for this to work. Ten replies is enough. Count the DMs, the YouTube comments, the thing three different people asked you at a conference. Repetition is the signal, not volume.

4. Revisit your own predictions

Any episode where you or a guest said "I think in two years" is a scheduled episode you have not put on the calendar. Go back, pull the tape, and score it out loud. What you got right, what you got wrong, and what turned out to be the wrong question.

Almost no show does this, and the reason is not courage. It is that finding the tape is annoying. Search your transcripts for "I predict", "in five years", "my bet is", "watch this space". Being wrong in public, with the receipts, buys you more credibility than being vague forever.

5. Ask every guest who else you should talk to

End every recording with the same question, once the pressure of being on mic is off: who works on this that I have probably never heard of? Guests answer that well. They know the people who do the work but never do press, and those people make better episodes than the names already circulating on the podcast tour.

Keep the answers in one list with the name of whoever suggested them. The referral is your opening line when you reach out, and it is the difference between a cold email and a booked recording.

6. The topic you circle but never land on

There is usually a subject that shows up in a third of your episodes as an aside and has never been the episode. A concept you use to explain other things. An example you reach for every time. A person you cite constantly.

That gap is the easiest hour you will ever record. You clearly know the material well enough to improvise about it under time pressure, and your audience has heard the trailer twenty times. To find these, you need frequency across the archive rather than memory, which is exactly what a topic list sorted by how often each subject appears will give you. The ones near the top with no episode of their own are your shortlist.

7. News and seasonal pegs, used sparingly

A timely episode gets a spike and then flatlines. An evergreen episode gets nothing on day one and keeps collecting listens for years. Both are worth making. The mistake is letting the news set your whole calendar, because a feed full of reaction episodes has no back catalog worth discovering.

A workable ratio is one pegged episode for every four or five that would still make sense to somebody pressing play in 2029. And when you do make the timely one, spend the opening on the durable part: how this kind of thing works, why it keeps happening. That is the bit people will still be sending each other next year.

8. The crossover

Trade an episode with a show whose audience overlaps yours without duplicating it. The version that fails is two hosts complimenting each other for an hour. The version that works starts from something you genuinely disagree about, or a subject where one of you has done the reporting and the other has the practitioner view.

Agree the specific question before you record. "Let us talk about our shows" produces nothing. "Is the thing we both cover actually getting worse" produces two episodes.

Where the ideas go

Ideas evaporate if they live in four places. Pick one list, and add a line for the source so you remember why it seemed good: the episode number, the listener who asked, the guest who suggested the name.

When you schedule, sort on two questions. Do I actually want to talk about this for an hour, and do I already have the material? Ideas that score high on both get a date. Ideas that score high on want and low on material become a research task instead of a booking. Ideas that score low on want get deleted, no matter how good the search volume looks.

---

Your archive, actually searchable

Tally imports your feed, transcribes every episode, and gives you a page per topic and per guest so you can see what your show has already covered and where it stopped short. It also runs live intelligence during a recording, which means the right information appears on screen while you are still talking. It is in invite-only beta and we are letting people in in waves. Leave your email and we will get to you.