Turn AI Into Your Personal Podcast Analyst!

Can AI actually help you make a better podcast?

This week, we break down how to use AI to analyze your podcast episodes, transcripts, titles, and back catalog so you can spot patterns, improve your content, and plan stronger future episodes.

Kevin shares how he used ChatGPT to review Buzzcast’s episode titles, downloads, durations, and transcripts to uncover what topics perform best, what kind of conversations create the most engagement, how each host contributes to the show, and where there are opportunities to improve. 

Along the way, we talk about why back catalog optimization matters, why transcript access is such a big unlock for podcasters, and how AI can be a genuinely useful tool for creators.

Steps to run your own podcast analysis:

1. Start with the data:

• Copy your episode list, including publish dates, titles, durations, and downloads
• Export or collect your podcast transcripts
• Upload both into an AI tool like ChatGPT inside a single project so it can reference everything together

2. Then give it some context: Tell the AI that older episodes naturally have more downloads because they’ve had more time to accumulate plays. Ask it to account for that before comparing old episodes to new ones.

3. Analysis prompts we used:

Broad analysis

• “Look at all the episode titles, durations, and downloads. What patterns do you see?”
• “Based on this data, what questions should I ask if I want to improve my podcast?”

Title analysis

• “Look at my podcast episode titles and downloads and tell me what title patterns perform best.”

Transcript and format analysis

• “Analyze my podcast transcripts and identify patterns in the conversations.”
• “What types of segments appear most often?”
• “Which segments create the best conversations?”

Host dynamics / speaking analysis

• “Analyze the speaking distribution across my podcast transcripts.”

Future episode planning

• “Based on my best performing episodes, suggest new episode ideas.”

Links mentioned in this episode:

• 6 Episode Title Tips for Podcast Growth
• How to Rank Higher and Get Discovered on Apple Podcasts
• 25+ Creative Podcast Marketing Ideas
• How to Choose the Right Monetization Strategy for Your Podcast
• How I Make Money Podcasting blog post

Contact Buzzcast

• Send us a text message
• Tweet us at @buzzcastpodcast, @albanbrooke, @kfinn, and @JordanPods

Thanks for listening and Keep Podcasting!

Transcript

Just A Little Episode Outline Chaos

Alban 0:00

Wait, what episode is this? There's two outlines in here.

Jordan 0:03

Yeah, Kevin's pulling like a 360 on us. Normally he doesn't even look at an outline. And then today, right before recording, he posts a second outline that is different from what I've been working on.

Alban 0:14

A 180, Jordan. I've watched enough snowboarding Olympics to know 360, you're facing the same direction.

Jordan 0:19

There's a turnaround. I don't know what it is.

Kevin 0:21

I mean, I have two modes. One, I can either not read the outline, or two, I have to just create my own outline from scratch with my hands. Like, I don't know that we should do my idea. I just thought my idea was our idea. I thought that's what we were doing. And obviously I'm wrong because there's three of us, and two of you were on the same page, and one of us was on a totally different page.

Jordan 0:41

I feel like we're not totally different pages. I don't know. You've you've got like a different version of the same book. It's just like a different translation. Um, let's figure this out.

Alban 0:50

Jordan, let's can we do this? You and I wrote the other outline with stuff that we were getting into. Let's do Kevin's outline. But then we have all this research and we just pull it in. It's going to be one episode, but it's in our minds. And so when it hits on this outline, that's what we go for.

Jordan 1:07

Yeah.

Kevin 1:08

All right. So what I hear you guys saying is if the episode is excellent, it's because we put all of our mindshare together.

Jordan 1:14

Yes.

Kevin 1:15

But if it's terrible, it's because we went with my outline that I posted four minutes before recording.

Jordan 1:19

And the listeners have to let us know what they think.

The Surprising Speaking Distribution

Kevin 1:22

Sounds good. Here we go. Okay, according to my outline, now is the time for us to do the cold open. Okay. And I thought it would be fun to open with this crazy statistic that I found when I was analyzing all of our data. And that is the speaking distribution for Buzzcast. Long time ago, Jordan used to actually publish in like the episode description in the show notes, like how much each person spoke. Do you remember that when we first started using Otter? Yeah. And I thought it was a really fun stat. My speaking portion was always the lowest. And I think Jordan was like in the middle, and Alban was always the most. That has changed radically. It must have changed like within the past year or so. But it is according to the research that I did. What does the research show? It shows that Alban and Kevin are almost identical. Jordan speaks the least. Here's what's really interesting, though, is that not only are we like almost identical, like we're like statistically impossibly close in our word count.

Alban 2:24

Yeah. So we have 1.8 million words across all of the podcast episodes that the three of us have been involved in. Oh, okay.

Jordan 2:33

So this isn't the entire BuzzCast back catalog. This is just everything.

Kevin 2:37

It's the entire BuzzProut back catalog that includes episodes where the three of us are on it. Gotcha.

Jordan 2:42

Okay.

Kevin 2:42

So Alban's right. 1.8 million total words spoken by the three of us in all episodes that we've appeared on.

Alban 2:48

And Kevin and I have spoken both 755,000. And for me, it's 897. Kevin, 398. So within 500 words spoken.

Jordan 3:01

That's wild.

Alban 3:02

I've still got a narrow lead, but I'm like that. Did you see the video of the marathon runner who won the marathon by 0.01 seconds? Yeah, yeah.

Kevin 3:10

Sprinted and caught the guy right on the finish line.

Alban 3:12

This is basically me throwing in a couple ums and ahs and taking the lead on the speech count.

Kevin 3:17

You're the blabbermouth. You can't you can't stop talking.

Jordan 3:20

That's crazy.

Kevin 3:21

Yeah. And so the research AI tool that I'm using said this is almost statistically impossible. Is it accurate? Do we know that? Yeah, I mean, I think it's just that I'm not, I'm not gonna go count the words myself.

Jordan 3:32

Yeah, I don't know how we're gonna test that.

Kevin 3:34

I I trust that it's not hallucinating on the word count. Like that's one thing computers are pretty good at is just counting things.

Jordan 3:39

I don't think it's accurate to real life because I know for a fact I've cut a lot of side tangents and things like that.

Kevin 3:46

But this was just fun and shocking. I feel like what the LLM wanted to write back to me was this is statistically impossible. Like there's no way that this could happen. But I think it had to put the word almost in there because we just proved that it can happen. It did happen. We have 200 and something episodes, probably like 170 something that we've all three done together. 1.8 million words, and there's 499 word difference between what Alban and I have said. Isn't that fun?

Jordan 4:09

Yeah. In reading this, what's more interesting to me is even though you two have a lion's share of the talking, we also are pretty darn equal on how many turns we get to speak.

Kevin 4:23

Yeah, I I like this term turns, which is how many times somebody comes in and says something. And sometimes when somebody breaks in, they might just say something short, like, oh, that's very interesting. I never thought about that way. And then somebody else comes in, that's another turn. But somebody, it seems like Alban and I, when we take our turns, we like to go on and on and on for a long time.

Alban 4:45

We grab a turn and we're like, all right, I'm going for it.

Jordan 4:47

We're letting go. Yeah.

Alban 4:49

Jordan gets a turn.

Kevin 4:50

He goes, Very interesting.

Jordan 4:52

That's cool, guys.

Kevin 4:53

Yeah. So our turn count is is all very similar. We're around 33% each, roughly. But yeah, the length of turns, there's a big differentiator. At least in words.

Alban 5:03

I mean, Jordan's probably got some like long laugh tracks in there. So she might actually get more airtime than the words are giving her credit for.

Jordan 5:12

My laughs are the fourth speaker of the podcast.

Exporting Your Stats and Transcripts

Kevin 5:15

All right. So what's the takeaway from this? The takeaway from this is that a lot of successful shows use this structure intentionally, and it feels like we do this naturally. So that's like a really good pattern that it helped identify and it encouraged us to lean more into this. Again, this is AI speaking, so you know, you can take a little grain of salt, but I felt like it was a good suggestion. Examples of shows that use this format are like Planet Money, Hard for Work, The VergeCast, all great shows. So it's like if we're doing something that they work hard to do, we just happen to do it naturally. I feel like, great, we'll lean into that more. We don't have it's not a big change, it's something that we naturally do. Um, but the current pattern that it identified was that Jordan would set something up, Kevin would give an explanation, Alban would give an explanation, and then we would move on to another subject. The improved flow, a couple of suggestions here. One that Jordan would maybe introduce a topic and then give a thought around that topic before Kevin and Alban came in with an explanation. And then later on in the outline, you'll see that um it even suggests there might be another opportunity for Jordan to take a turn to summarize the two uh points or counterpoints that were made before moving on to another topic. And what that would actually do is probably give Jordan more turns at the end of the day. So increasing the number of turns that Jordan has and balancing her word count probably a little bit more with Alban and Ice. So why did you do this experiment, Kevin? Uh because I I thought you guys told me to. That's why.

Alban 6:36

This is this is like parenting right now.

Kevin 6:40

I only did it because you told me to. Like what? When did I say this? So, I mean, we did a quick cast last week where we talked about optimizing episode titles because we thought in the world of podcast optimization, an easy place for everyone to start would be by optimizing their episode titles. Yeah. Right. And while just doing a little bit of research on that, I discovered that we're sitting on this massive treasure trove of data. And Alban, you had the same epiphany, right? When you started pushing into this, massive treasure trove of data, yes. And in the past, if you had a massive amount of data, it required a massive amount of work to be able to analyze that data.

Jordan 7:17

Yeah.

Kevin 7:17

Well, it's a whole new world. AI is fun for a lot of things, making silly pictures and putting cowboy hats on people and stuff, sure. But it is also really good at analyzing huge amounts of data very quickly, finding patterns and making suggestions and all this other stuff they can do. So it started with simply just going to the episodes page within Buzzsprout and literally highlighting all of our episodes from top to bottom. And when you highlight all the data on that page, you get published date, you get episode title, episode duration, and download count. I highlighted it from top to bottom, copied it, pasted it into Chat GPT, and I did a little bit of explanation so it didn't get the data wrong. I said, like the first line is the episode publish date, second line is the episode title, third is duration, and fourth is the number of downloads. Based on this, what patterns do you see? And it gave me some insights back. And then I was like, well, well, wait a second. It's making some assumptions that might not be good assumptions because it just assumes everything's equal. So older episodes have a lot more downloads because they're in your back catalog. They've been live for you know a year or more. And all our back catalog episodes get three or four downloads a day. They don't do huge numbers, but like over time that accumulates. So it made it started to look like all of our old episodes were the most popular. And so then I said, hang on a second. New episodes don't have the benefit of history or a long time duration on their side. So you have to figure out some way to account for that while you're evaluating what's working and what's not. And so I typed that into chat and then I opened up the little console next to it and I saw all the math that it was doing. I don't know. It's like a like a beautiful mind type thing. Like numbers are like flying through the screen and it's creating all these things and it's trying to figure out how to account for time. And so then it started chunking everything. So it said, okay, well, I'm gonna evaluate like this block of episodes together and this block of episodes together, and this, and then I'm gonna weight them differently based on time. And I was like, oh, this is so amazing. And then it figured out how to weight them appropriately, and then it said, Oh, this episode was actually just published three weeks ago, but it's actually just as good, if not better, than this episode that was published two years ago that looks like it has twice the number of downloads or three times the number of downloads. And I was like, that is something that you cannot see by just looking at your stats alone in BuzzSprout. This is a new unlock. This is powerful. And so once I had it to that point, then I wanted to get more context in there. So I want you to know more about my episodes than just the published date, the title, and how long it is and how many downloads it had. I have the full episode transcripts for all of these. How do I get all of that information into the data set? So now you know not just the metadata around an episode, but actually you know the full content of every episode too. And then I remembered in BuzzSprout, we have this awesome backup feature.

Jordan 9:54

Oh.

Kevin 9:54

So if you go into your Buzzsprout account, you go to your podcast settings and you go to backup and you scroll to the bottom, there's a link where you can download all the transcripts for every episode in one large text file.

Alban 10:06

You know in uh Interstellar where he's like looking through the uh bookshelf and he's seeing himself do stuff in the past. Yeah. And he's like, no, stop. That's Tom yelling through the bookshelf right now, like, stop telling people to export all their data to get the transcripts.

Kevin 10:20

No, the transcripts, that's not bad. It's not bad.

Alban 10:24

I believe.

Jordan 10:25

It's just books start falling off the shelf.

Alban 10:28

Kevin's like, what are these books falling off for? Oh, I guess I'll just like go to Mars or whatever he does and leave my kid. Tom's like, don't the transcript. Just don't tell everyone to export their data.

Kevin 10:39

Yeah, that's fine. That's a good point. But this alone will not crash the server. That's actually not a super intensive call. That happens real time. It's not a report that you have to run and wait on. It's just a huge text file. And mine ended up being like 10 megs, which is pretty big for a text file. It is a huge text file. But then I have that sitting on my desktop. So then I drag that into Chat GPT and I'm like, here's the full transcripts for every episode that we've ever done. Like commit this to memory, and then let's start like re-evaluating patterns and stuff like that because I want to be a better podcaster.

Jordan 11:09

Yeah.

Kevin 11:09

And so that is the data set. I was like, connect these two things. Connect some basic stats data, recognize that the newer episodes don't have the benefit of a long published history on their side. So account for that. And then here's the full content of every episode, every word that we've ever spoken.

Jordan 11:27

I was surprised that you copied and pasted the episode page with the downloads and stuff, exporting the stats CSV from the stats page.

Kevin 11:36

Yes. Well, here's the thing. I was just like experimenting. And I was like, what's what's the easiest way for me just to play with this data? Because one of the things that can be intimidating is, oh, I have to go through this process of getting all the data and making sure it's all in the same format. How do I do that? Do I do it in Google Sheets? Do I do it in this? And I was like, I don't, I'm not going to mess with any of that because I want to keep this as simple as possible for people. And so I'm going to upload like the messiest data possible and see if the LLMs are good enough to put it all together. And sure enough, just like with a few words, I was like, here's the episodes and here's the downloads and the published dates. Here's a completely separate file with just all the transcript data. Just match them up by the title. And it was like, no problem. I can do that. And I was like, great. I don't know how to like discount for newer episodes, don't have the benefit of long published duration on their side. Can you account for that? And it was like, sure. Here's a ton of math that I'm doing, and I'll figure out how to weight them appropriately.

Alban 12:25

I was like, great. What were you throwing this into, Kevin? This is like 5.4 Pro or something. You're doing this much work?

Kevin 12:32

Um, it was it's it's ChatGPT 5.4 and I set it to auto. When I set it to pro, it's just taking too long. And so I did run a few queries on pro, but it was taking too long. And I was like, I don't think the insight's that much better. Because this really isn't super complicated stuff. So I don't need you thinking about like what patterns can you identify for 10 minutes. Like it was almost the same patterns as when I just left it on auto and it would run a lot faster.

Alban 12:54

So to grab all this data, obviously, if you have this transcribed, it's best. So you use Adobe Podcast, you use Riverside, you use Descript, all those have it. Buzzsprout, we create them. Can I share something new? Yeah. This is breaking news. We need a breaking news like button. Why is it that? What does breaking news did it like historically come through uh Morse code or something? Telegrams. Yeah. It's like the AP out in California is now like buzzing in and letting us know.

Jordan 13:23

I think it's just a noise that like meant urgent, like this is an important message.

Kevin 13:28

Okay. Yeah. Well, before you break the news, Alban, let me tell you that during our conversation when it was telling me how to become a better podcaster, it said lead with the benefit. Don't lead with the feature. Like, don't say, oh, we have this new feature, and and here's why, you know, we did it or whatever. Lead with the benefit.

Alban 13:44

And that's why this segment was so good. Because what we talked about was the benefits of analyzing your transcripts. But what if I don't have transcripts? Oh, I can't analyze the data. Well, now there's a brand new feature benefit. And that is that we are going back and we are transcribing everything across all of the new Buzzbrout plans. Woo! We are going back and transcribing every episode that we have. Our transcriptions are so much more accurate than they used to be. There's some really, really nice improvements to the transcript editor. There's some really nice ways now to create chapter markers in the transcript editor. So if you're going through and reading and you don't love the transcripts that co-host AI came up with, you just put them in yourself. All that is coming out. Some of it came out today, some of it came out yesterday. This is going to be all out for everybody by the time you listen to this episode. And now you're going to have all these transcripts, and they're much more accurate and they're going to be useful for analyzing your own data.

Jordan 14:47

So our podcasters can go back and edit the transcripts. They can take a look at it and mess it around in the editor and everything for all of their episodes, even if they don't have co-host AI.

Kevin 14:57

Right. So this is for people who are on the new plans. If you're on a grandfather plan, we're not forcing you to move to the new plans, but the new plans are a little bit more expensive. But at the end of the day, it might end up being cheaper. So if you're like on a $12 plan and you're paying $6 for co-host, that's $18 a month that you're paying. The new plans include transcripts for free. So if really all you're using co-host for is the transcripts, you can jump to the new plan. You get more hours instead of three hours, you get six hours. So you get twice the number of hours and you get transcripts included for free. And we'll transcribe your whole back catalog for you.

Jordan 15:29

That's awesome.

Finding Patterns In Past Episodes

Kevin 15:30

And so we think we've really wanted to do that because of analysis like this. When you have transcripts for every episode, you can download that full transcript file. You can drop it into an LLM, and then you can ask it analytical questions like I was asking it for the purpose of this episode, which is basically how do we become better podcasters? What's working? What's not? What can we get better at? Look at old episodes, break them. I'll get into more fun queries that we could do. But you can also ask it silly questions. Like I just wanted to ask it, like based on everything that I've ever said, what do I think about like why don't we publish Buzzcast on Spotify? And do you think I've been too harsh of a stance? Or do you feel like it was appropriate? What's Alban's?

Alban 16:08

I mean, knowing knowing how these LLMs work, it was like, oh no, Kevin, you need to trust yourself.

Jordan 16:13

You're perfect. Jordan and Alban are out to get you.

Kevin 16:16

Yeah, yeah. You can direct that. I was like, you know, give it, give me a candid take. Like, give me, give me an honest opinion. What would other people say? Yeah. I would say, like, what would Jordan say? Would she agree with me? Like, you can just have fun conversations about stuff that you've already said. And those are just for fun. But in order to do that kind of stuff, you have to be able to have full transcripts of all of the content you've ever published. And what's what's really exciting is it just gets better and better over time. So we have 200 and how many? 218, 219 episodes that we've published.

Jordan 16:42

Yeah.

Kevin 16:43

So there's a lot of stuff there. Uh, if you just have 10, it's still going to be decent. Like it's good. But what gets exciting is like you have content like when you hit episode 50 and you want to do something cool, like this is a great show topic. I've got 50 episodes now. I went back through all the things and like what were the 10 most memorable moments from my podcast in the first 50 episodes or first hundred episodes, and it can help you come up with these ideas.

Jordan 17:03

Yeah.

Kevin 17:04

And so, in order to unlock all of that stuff, it's important to be able to have full transcript data. And so that's why we wanted to give that to as many people as possible, make it as affordable as possible.

Alban 17:13

Yeah, I really love this. I think as soon as you have transcripts, it opens up a world of new things you can do. I think maybe honestly, I feel like the script kind of invented this when they did their transcript level editing, like they transcribed the audio and then they said, okay, now you can edit that, which edits the final episode. It was such a unlock for editors. And then every LLM is text in, text out. It's made it really easy to analyze data once we have the transcripts. So what's the first prompt, Kevin? What was the first thing you asked?

Kevin 17:47

Okay, so we have all the data. It might be fun depending on the AI tool you're using. I created a project for this, so I can come back to it at any time. I don't have to like re-upload all the data. When you create a project in these AI tools, it remembers that data set.

Jordan 18:01

Yeah, you can upload them as sources.

Kevin 18:02

Yes. And it doesn't usually allow you to um, well, it might remember like chat history, things that you've discussed. It doesn't remember files and stuff that you've uploaded across different chats. So my recommendation would be to create a project or some sort of grouping, depending on what AI tool you're using. Great question, Alban. The first prompt after all this data was loaded in, I said, look at all the episode titles, durations, and downloads, and what patterns do you see? So super broad, super general. What do you see? Like I don't even know what to ask yet. This is sort of me asking the question of what question should I ask if I want to become a better podcaster, if I want to improve my show.

Jordan 18:38

Oh, yeah.

Kevin 18:39

Help get me started down the track. And so that's what I came up with. Look at all the episode titles, durations, and downloads. What patterns do you see?

Jordan 18:47

I like this because most podcasters don't even know what the right question is to ask. I mean, it's kind of like you don't know what you don't know, right? Yeah.

Alban 18:54

This kind of reminds me of uh some of the podcast audit services we've seen over the years. Somebody who's like set themselves up as a coach will say, Hey, you know, send me your podcast and a thousand dollars, and I'll go through and listen to a bunch and give you a bunch of pointers. And you're getting a lot of that same type of feedback here, Kevin.

Jordan 19:12

Yeah.

Alban 19:12

Yeah.

Jordan 19:13

So what kind of things did it show you as patterns that it detected?

Kevin 19:17

The first stuff it came back with was like it started to categorize all the episodes that we've done.

Jordan 19:21

Oh, okay.

Kevin 19:22

So it came back with five different general categories that it it put all the stuff that we talked about into, which is like tactical podcasting tips, industry news, creator economics, which would probably be like all our monetization talk and advertising and all that kind of stuff.

Jordan 19:36

Yes, I was thinking, yeah.

Kevin 19:38

Uh event recaps, which I'm surprised we did enough event recaps for it to like call that out as a category. Yeah. It was also like a red flag, like we're recapping podcasting events too often. And then platform drama.

Jordan 19:52

Kevin, I think that's all you.

Kevin 19:54

I'm the yeah, I'm the big drama mama. We need we just need a new segment called platform drama. Right. And so after it kind of created those five main categories, it also pulled out a couple of episodes about podcast strategy and mistakes performed much better. And examples were like podcast marketing ideas. That one we knew. We like we knew that was a banger episode. We got a lot of interaction on it. A lot of fan mail came in around that. Listener mistakes, which I know we've done a few segments specifically around like podcasting myths and stuff like that. It probably threw things like that in there.

Jordan 20:29

I feel like those are always a good hook too.

Kevin 20:32

Yeah, those are a good hook. But then also I think like a lot of people who write in with fan mail with things like they're feeling uneasy about or insecure about, and we're just we try to be encouraging and say, hey, we all have this feeling. We all hate the sound of our voice the first time we hear it recorded and all that kind of stuff.

Jordan 20:45

Yeah. And we also did that like uh review episode. So this might fall into that bucket as well.

Kevin 20:51

Yes. We did a whole episode where we kind of gave feedback on a bunch of different podcast episodes. And then podcast analytics, which of course we talk about all the time.

Jordan 20:59

Like today.

Kevin 21:00

Yeah. And so it said those episodes specifically are answering the question of how do I make my show better?

Alban 21:05

All right. So that whole bit was analyze my all my data. What do you see standing out? What topics are doing well, titles, durations, downloads, what patterns. And it came up with here's where you talk about, and then here are the ones that seem to do the best. Right.

Jordan 21:19

What's really interesting about it saying, okay, your most popular episodes answer this one core question, how do I make my show better? That can actually inform future episode planning as well. When you're trying to figure out what kind of topics you want to cover, if you know the core question that your audience is asking, it's a lot easier to say, Am I answering this? Am I serving my audience? Am I giving them the content that they want?

Alban 21:42

Yeah. Yeah. This was in Jordan's in my outline, Kevin. That when you analyze your data, you get two things. One, you get forward looking. Here are some prescriptive ideas. You know, now we've come up with some new patterns we had never thought of before. Here's how we can change. The show. But you also get a bit of ideas for optimizing the previous content. Because if there's old stuff that you know, oh, that actually hit these areas that are really exciting and yet an episode didn't perform well, it may just need like a title update or some sort of optimization. So you have ideas for uh the back catalog as well.

Jordan 22:22

I like that.

Kevin 22:22

That's a really good point.

Jordan 22:23

Yeah.

Kevin 22:24

So from this segment, I really feel like the big takeaway is that when you ask these general categorization questions and like look at episodes that perform better versus worse and what sort of patterns do you see? Like the first one or two that come out, I think are the ones that did not surprise me at all. So again, it it it it put this into five categories for me for us. Tactical podcast tip and industry news were the first two. And then the next one was a little bit more eye-opening, you know, like creator economics. I'm like, do we really talk about monetization and monetization strategies and stuff like that that much? Event recaps. My eyes are open even further. Like, oh my gosh, I think that's too much event recapping. That it's it's its own category. And and platform drama. I know we go on side quests sometimes and we get wrapped up in some of the stuff, but those aren't like the real important things for us to be talking about. Yeah. And yet we are spending a decent amount of time talking about that stuff. Like we're trying to help people keep podcasting, become better podcasters, pull joy out of the something that they're they're stepping into for a hobby and stuff. That's the main core of our episode. And it's kind of only like the top two or maybe three categories that are doing that really well. The other two aren't. And the same thing when we talk about the episodes that perform the best, the podcast marketing ideas, not a surprise there. Listener mistakes, again, maybe like that's a little surprising. Podcast analytics, like knowing that that is a really high-performing topic for us to cover. We're doing a whole episode on a new, fun, exciting way to analyze all your podcast data.

Title Optimization And Back Catalog Value

Jordan 23:50

Yeah, it's completely different than what we've done in the past. So this next segment, title analysis. We kind of touched on this last episode, right?

Kevin 23:58

We did. So we don't need to go into a ton of depth like we did in the quick cast. Uh, if you didn't happen to catch the quick cast last week, we did what started off on this journey. We started in the like the podcast optimization, like the podcast search, what do we call it? It's not SEO, it's not search engine optimization, it's podcast app optimization. That's a good one. P A O.

Alban 24:18

Pow Payo. Yeah. Um, so in my outline, Kevin, the deleted outline, I found this thing from Triton. They went and did all this analysis, and one of the things they found was that back catalog downloads, which they categorized as episodes that have been published more than 12 weeks earlier, 12 weeks old, still made up a substantial amount of listening. And so news, old content isn't as important. So only 20% of content comes from things that are over 12 weeks old. But as soon as you get to things like science, comedy, society, culture, like 40% of all downloads are the back catalog, all the way up to health and fitness, which is about half of all downloads are coming from content that's over 12 weeks old. And the takeaway I took from this was back catalog is really, really important for a lot of podcasters, even news, which I would have said was almost like just don't even make them visible. Who cares? No one's going back and listening to what happened in the news six months ago. No, it actually is still 20% of all downloads. What they showed, I think this might have come from Dan Meisner when he used to write for Pacific content, that that back catalog is so valuable because those are all entry points to the podcast. If 40% of all your downloads are up to 50% are coming from the back catalog, each of those is a potential first episode.

Jordan 25:49

Yeah.

Alban 25:49

And so once you get some real data around what type of titles work for you, what type of content works for you, if you have opportunities to go back and optimize that back catalog, that's not a small opportunity. One, someone might find it in Google or in some AI tool. But also, like once they find that they like your show and they're listening to it, they're going to scroll the back catalog. And those titles need to be descriptive enough that somebody could see it and go, Oh, I do want to listen to that one. Download, download.

Kevin 26:22

You know, I never thought about it like that. So he's saying, and you're saying, that their first episode listens because the the chances that they stumbled upon that episode was as the result of some sort of exploration, whether they searched in a podcast listening app or they searched on Google or something, somehow they landed on an older back catalog episode. And so the point that you're making is it's important to have that the title of that episode optimized for terms that people would be searching.

Alban 26:49

I think I'm combining three things here. Okay. This report from Triton, back catalogs are really important. This article from Dan Meisner, this basically said the best shows actually have really big back catalogs. And some of that is because every one of those episodes is a potential first. And then I've got some of my own analysis, which is you can't really know how to write a great title for your podcast until you've done a few episodes and then you can do some analysis to figure out what works. You'll know what topics are more attractive. And once you know that, now go optimize that back catalog. Right. And we're not just optimizing for AI or SEO. We're optimizing for your super fan who scrolls that list. I've done this. I mean, imagine someone who goes and listens to one episode of Joe Rogan and they like it, they scroll back through the back catalog of, I don't know, 4,000 episodes and they're all crazy names. They just go, okay, I don't know, and they move on. But if there's a handful of episodes that are called out as like, these are the top ones, go listen to these, and they have really good titles. You're gonna get many more opportunities to get new downloads. And up to half of downloads are coming from that back catalog. This is a big, big opportunity. Right.

Jordan 28:01

Everyone knows that podcasting is a long game, it's a marathon. And I think that this really supports that. And it's not just in order to grow your show, you have to like keep podcasting for a long period of time and build up that back catalog. It also has this element that you need to optimize that back catalog so that you can make that back catalog more discoverable and really put your focus on it instead of just, you know, publishing an episode and moving on, never looking back at it again. It's good to revisit those kinds of things.

Kevin 28:29

This brings me back. I I said this earlier, but I might have not covered it in the depth that it it warrants. And that is, I said something to the effect of when you just copy and paste all your entire episode history into an LLM, you have the titles and the published date and the like the number of downloads from top to bottom. And your newer episodes might have whatever, 20 downloads because you just launched them yesterday, and your older episodes have a couple hundred downloads. This assumption that could happen from the LLM that, oh, these older episodes are a lot more popular. You must be doing something terribly wrong with these new episodes. You have to tell it. You can't assume that it knows how podcasting works. Okay. So I want to be clear with the very simple instruction that I wrote that led to the LLM going on a four-minute thinking journey and spitting out tons of Python code and running into errors and figuring out how to wait episodes appropriately based on how long that they've been published. Because it might just sound like, oh, well, Kevin's a programmer. He figured out how to do that. I can't figure out how to do that. No, you totally can because this is what I wrote word for word. I just pulled up my history. Please keep in mind that newer episodes naturally have fewer downloads as episodes accumulate downloads over time. Bat catalogs will continue to accumulate a few downloads every day, even if they aren't very strong. That's all I wrote. After I wrote that in, it thought for four minutes and 17 seconds and says, like, yes, you're right to push on that. And then it told me what it did. And it came up with a massively complex way of weighting older episodes versus newer episodes. And then the data that it was giving me back was much more appropriate. And to a non-programmer, that is something that probably 12 months ago, 24 months ago, for sure, 36 months ago, absolutely, was impossible. We just did not have these tools available to us as non-programmers. Um, you would have not only had to have been a programmer, but you had to have been a darn good programmer and a good mathematician and everything to be able to do analysis like this. Now, anybody with a $20 chat GPT subscription or even a free model on a lot of these services can run those sort of very complex equations and analysis on your data set by just telling it, hey, don't look at old episodes and new episodes the same way because old episodes can accumulate downloads over time. Can you can you figure out a way to weight them appropriately? And that's all you have to say, and then boom, it figures out all the math to do it. So let me pull us back real quick. We just want to summarize this because again, we did cover it in more depth in the last quick cast. But uh the patterns that recognize analyzing the titles for BuzzCast specifically were pattern one, clear practical titles perform best. So examples of that were like podcast marketing ideas, podcast listener mistakes, et cetera. And the answer to the question of like why does it look like they perform the best was because they promise immediate value. It also suggested that lists outperform vague titles. So we talked about this, like episode titles with numbers in them seem to be performing well. At least for now, that might change over time as people start doing like listicles type type stuff. If anybody starts abusing that, it could change. But at least for now, numbers and titles for podcast episodes seem to work. So, like the five mistakes of yada yada yada or 25 ideas, two, whatever, those still perform well in the podcasting world. And then number three, podcast titles work when the impact is clear. So things like changes to Spotify that you should know about. Like it's a there's a clear impact here, and we're saying podcasters need to pay attention to this or Apple podcast updates, again, like something that you should know about. And because the title is telling them like not only what changed, but why they should care. So I think that's as as deep as we need to go on titles. Again, we covered it in depth last week.

Alban 31:58

Could you give us the exact prompt that you use, Kevin, for people who are going to do this themselves?

Kevin 32:03

Yeah, that was again very simple, which is look at my podcast episode titles and downloads and what title patterns perform best. So, Jordan, you might be able to copy and paste some of these prompts into our show notes. Do you think that's going to be helpful to people?

Jordan 32:17

Absolutely.

Kevin 32:18

But none of this is rocket science. I tried to keep it as simple as possible because, first of all, I'm not the world's like greatest AI prompting podcast expert. I'm not billing myself as that. I'm just, I'm just asking the simplest questions that I can think of. And if I can't think of a question, then I simply say, I don't know what to ask and ask. Give me some ideas of questions that I could ask.

Transcript Insights On Segments And Roles

Jordan 32:37

Yeah. Your next analysis here, and we kind of touched on this a little bit, was the transcript analysis, but you are approaching it a little bit differently here, where you're asking AI to analyze all buzzcast transcripts and identify patterns in the conversations. And so I want to go over what podcasters can glean when they ask it to look at the transcripts and identify patterns, like what kind of things can get from that.

Kevin 33:01

I thought this was one of the most interesting things because now it's started to try to figure out where our engagement was happening. And so, of course, it was continuing to look at the download numbers. Like that's the main key that it had to figure out like what's working and what's not. But at this point in the conversation with the AI, it also recognized that we do a fan mail segment every show. And so we ask questions and we get listener feedback through fan mail. And so it started to pick up on this pattern of oh, another signal that this was a very popular episode is how much fan mail you get in the next episode.

Jordan 33:37

Oh.

Kevin 33:38

Okay. So then it's telling me this. And I was like, oh, that's really smart of you to recognize that. And sometimes it's people answering questions directly. Sometimes it's just engagement in general.

Jordan 33:48

Yeah.

Kevin 33:48

And so maybe your episode only had, you know, like maybe it was very average in the number of downloads that it got, but on the next episode, we had a bunch of fan mail. And so then it would say, I'm going to now mark this previous episode as like a high engagement episode because it generated 10 fan mail responses, even though the downloads, just looking at the download number itself, didn't elevate it as an exceptional episode. And so the segment types that it listed and started noting as signals of a really uh good episode. That number one, the listener feedback from fan mail and stuff, we started to have conversations around that. But it said the strongest episodes seem to combine multiple types of things. And so, like when we have an episode, again, all of them have fan mail. So it kind of just put that as like table stakes. They all have fan mail. But then also when we cover an industry topic, we give some tactical advice, and then we also have some deeper discussion around that, sort of like editorialized content. When we combine all four of those things in a single episode, those end up being our strongest episodes. I did ask a bunch of questions about our shorter quick cast episodes versus our full length uh multi-topic episodes. And it was it was very clear from the data and the patterns and the analysis that it performed that they're they're almost on equal footing.

Jordan 35:01

Oh.

Kevin 35:02

Which is good because like if somebody follows the show or subscribes to it, it seems that they're listening and engaging just as much with a quick cast as they are with a full-length episode.

Jordan 35:11

Yeah.

Kevin 35:12

So what this is mostly pointing to is it's not about how many topics we cover. So in a full episode, usually we're covering more than one topic. It's that when you do cover a topic, make sure you cover it from these three different angles, which is industry topic, like we give an overview of what the topic is. We give tactical advice around that topic, and then we also have a slightly deeper discussion around it. And that can all be scaled up or down depending on the length of the episode. And so a long episode typically covers multiple topics. That doesn't mean that that's going to be better. It's when when I say the strongest episodes combine multiple types, it's multiple types of discussion around topics, but it doesn't necessarily mean you have to have multiple topics. But the takeaway from all of this is that it creates insight, community, and practical advice. And then I have a few example prompts that if you want to get this type of insight for your podcast, that you could type in exactly word for word, which is again, your transcript data has been loaded in already. So you would just say, analyze my podcast transcripts, and then you'd say like one or two or a variation of one of these questions, which is like what type of segments appear most often, or which ones create the best conversations. So again, you'd modify that if you have a solo podcast versus if you have uh guests on your show, or if you have a consistent co-host or co-hosts, multiple co-hosts, you can modify these prompts, but generally you're just trying to say like what's working and what's not, like segment-wise. And if I want to lean more into what's working, what's what's a format that I should look to follow? This was just started as a fun exploration. I was like, based on our transcript data, like, who do you think Jordan is? Who do you think Kevin is? Who do you think Alban is? And we were just, I was just like just seeing what I could get it to spit out. And I know you two both very well personally. And so I was like, oh, yeah, like some of that's accurate, some of it's not. But then I tried to get a little bit more practical with it. And I was like, so like in the eyes of the listener, like what role do each of us play on the podcast? Or like what are the archetypes that we fall into?

Jordan 37:08

Yeah.

Kevin 37:08

Who are we?

Alban 37:09

I found it interesting. It said that Jordan is the listener's voice. She does transitions and reactions. Yep. Kevin does insight and explanation and album analysis and technical debt. Yeah. I don't think of myself as technical debt. I'm on the opposite side of that.

Kevin 37:25

I was with you. I was surprised that it thought that your explanations were more technical. But then, like looking at some of the segments, and I didn't paste all the segments that it pulled out as examples of this, because anytime I didn't understand some of the analysis, the conclusions it was drawing or patterns it was drawing, I would ask for clarification. I'd be like, so you say Alban is speaking more technically. Can you give me examples of this? And it did spit out three or four or five, as many as I wanted, it would keep going of explanations of where you're going, like more on the technical side of things, like more analytical. And I'm a little bit more like philosophical. It sort of identified that. It was like Kevin is the one who makes the principal stand against Spotify, like listing our podcast in Spotify, while Alban wants to explain, like practically, like how is this actually playing out?

Jordan 38:08

I would agree with that. And I think being the listener side of things is probably due to also being the producer. And so I'm sort of listening to your conversations, how our listeners are going to listen to it. I'm listening for cues where we need to clarify some points and stuff like that. So I agree with that.

Alban 38:25

That came out of our last episode that Jordan and I did about editing. One of your takeaways was listen to the answers. And if you don't like the answer, just ask the question again in a different way. And then you'll get the answer that you'll actually leave in. So that uh you don't have to, what did you call it?

Kevin 38:40

Frank and bite the episode.

Jordan 38:41

Frank and bite, yeah.

Kevin 38:42

Oh my gosh. So I wasn't on that episode, but I listened to it, and I think you guys definitely set some sort of world record. I mean, Frankenbite is a word that I think Jordan invented in the first place, right?

Jordan 38:52

No.

Kevin 38:53

Oh, you didn't. Okay. I still don't believe that's a real word. If you guys go back and listen, I'm sure Jordan knows because you edited the episode, but the number of times you guys said the word frankenbite was the chain. Is that a good thing? So much.

Jordan 39:07

No, it's not a good thing. We beat that word to death.

Kevin 39:11

And so that was all fun and exploration stuff, but it kind of boiled down to, again, so for listeners of this show, the biggest theme that it could put around Jordan was that you provide the listener perspective. Like that's your role is to keep us on track and like what do listeners want to hear and what questions might be going through their heads right now. And you interject that into the conversation when appropriate.

Jordan 39:30

Yeah.

Kevin 39:31

The role it gave for me was big ideas. I don't know why I said it with that inflection, but big ideas, guy. Maybe just blowing things out of proportion. I have no idea. And Alban, again, the surprising thing was the technical analysis. Yeah. Because Alban kind of getting into the nuts and bolts of how things work, how the sausage is made, sharing maybe more of the backstory of the different iterations that we went to before we came to this conclusion or built this piece of software, things like that. Again, like important things because if you get this type of feedback around your show, I think it's either going to resonate with you and feel good, or it's going to feel off. And if it feels off, it gives you something to work on, like in terms of changing that. Yes. But if it feels on, it gives you something to lean into.

Alban 40:12

Have you ever heard the quote, what gets measured gets done. Yeah. When you measure something and you report it, there's this whole theory of like problem solving that all it is is like once you identify the problem, you've done 80% of the work. As soon as Jordan started posting, Alban spoke 52% of this episode. I'd never thought, oh, I'm dominating the conversation. Once I saw it, Jordan didn't say it was bad. No one gave me any direction. I just looked at it and went, my aspiration is not to talk over half of this episode. I would like to be more even. And reading some of the transcripts, I went, I can be long-winded. I'd like to tighten that up. And then this made it look like that actually has changed a bit since Jordan used to post those. So when AI reveals things like who's talking a lot? Are there pacing problems? Are there segments that are running too long? What role is everybody leaning into? Some of those, like Kevin said, will show up as, oh, that's exactly right. You got us. Like when you uh read a horoscope and you go, nah, that's me for sure. Yeah. And sometimes you'll get a bit of feedback that can just feel wrong. And it's not because the feedback is incorrect, it's because, oh, that's not the kind of person I want to be. That's not the kind of podcaster I want to be. That's not the role I want to play on this show. Yeah. Yeah. I don't want to play the role of someone who talks 60% of the time.

Jordan 41:32

Yeah.

Alban 41:32

And then you're just going to naturally move in the direction that you want, you know, to show up as. And I think you can become a better podcaster. So as soon as we start to identify it and get some feedback, even with a grain of salt, I think it ends up making us better at what we do. Yeah.

Jordan 41:47

Yeah. And anything you can do to become a better co-host or a better moderator of the conversation. It's funny. That's not really something that I would have thought we could glean from AI analyzing transcripts is how can I improve my speaking part on a podcast, right? Like my contributions to this podcast. Yeah.

Kevin 42:06

It's interesting. Well, and that was a huge takeaway for me is that when I looked at this, I realized not only does it it feels right, it feels accurate. And we had a bit of a back and forth before we kind of got to this place, but it also reminded me that you know what? I have a role to play here. And if I don't do it, it's not Alban and Jordan's natural bent to do it. So if I don't like zoom out and try to find the big idea here, the big picture, how this affects podcasting like today and maybe long term, that's not either one of yours natural bent to do it. So it's important that I remember what I bring to the table and that I do it. I fulfill my obligation there. The same thing like with Jordan, with like a listener perspective. Alban and I could ramble on and on about a whole bunch of like technical stuff or why we're doing something, but neither one of us are going to have like top of mind, like how does this impact the everyday podcaster? Or what would an everyday podcaster ask if they were a part of this conversation right now? And the same thing with Jordan and I. We could go on and on about some, I don't know, editorializing some decision that Spotify made or something like that. And Alban might be like, oh, well, there's actually some practical reasons too why we might not want to implement that technology. And let me tell you how involved like that might actually be and how complicated it could make the UI. And so he brings that, you know, sort of deeper practical technical analysis to the conversation. That's where his mind goes. So we all have these roles to play. And it's important for us to remember sometimes it just feels like I don't have a lot to contribute, or that feels petty. But we're balancing each other out. We're all bringing these perspectives. And so remembering those, keeping those top of mind during the conversation and saying, you know, if if I don't bring it, neither one of these people are going to bring it because they lean a different direction. Is helpful.

Alban 43:41

So, Kevin, do you have a prompt or what did you use to get this information?

Using AI To Plan Future Topics

Kevin 43:45

This one was probably the simplest of all of them, which was analyze the speaking distributions across my podcast transcripts. That's where we started. And then it spit out a whole bunch of information. And then I started asking follow up questions from there. But I feel like that's I just want to give you the starting prompt because you Your conversations might go all different directions depending on the patterns that it finds. But the the key words here are speaking distributions. So analyze speaking distribution across my podcast transcripts and then have fun from there. Okay.

Alban 44:13

So so far we've done a lot of what's happened in the past. What titles perform well? Who's talking when? What segments do we do? Blah, blah, blah, blah, blah. Can we shift now towards the like forward thinking piece? How do we use this to plan better episodes? Some of it will happen naturally, but is there anything explicit we can do so that the future content is better?

Kevin 44:36

Yes, 100%. There's a reason that I put this more towards the end of the outline than towards the beginning. And that is because you want to make sure you're comfortable enough in the session that you're having with the AI and the data analysis that you've you've gotten through the like hallucination part. Because now we're asking it to like help us predict future stuff. This is where hallucinations can start to appear. And so we want to make sure that we've gotten comfortable with the responses that we're getting from the data set that we provided. And we've like we've already gone through the exercise of saying, hey, new episodes aren't going to have as high download numbers as old episodes. And we've started to look at title suggestions and how titles could impact downloads. And so some of these episodes that look weak, maybe they weren't really weren't weak. They just had a bad title. Like we've already worked through all of that stuff in our session to this point. And so now I felt comfortable saying, how can you help me plan for some future content now that you have a better understanding of what really is a high performing episode versus a low performing episode? Does that make sense to kind of set the groundwork?

Jordan 45:41

Yeah.

Kevin 45:41

In the in the analysis tool that you're using first before you start asking it to predict the future for you?

Alban 45:46

Yeah, because you want to make sure that the baseline analysis is grounded in reality. Yeah.

Jordan 45:51

Yeah.

Alban 45:52

And you want to be grounded in reality so that when it goes off, you can better guide it back and notice like that doesn't sound right based on what I've seen from this data. And I've kind of been checking the whole time. So there's a story in the news recently. A guy is, I don't know if he's suing open AI or what, but he I guess had a chat that went on like millions of words with Chat GPT about quitting his job to join the PGA tour. Okay. Like professional golfer. I can relate right now. He is not a professional golfer. He just was like posting stuff about golf, and it was like, oh yeah, this sounds good. And he's like, I kept asking, like, are you sure I should quit my job? Are you sure this is real? And it's like, yes, this is such an amazing opportunity. You got to go for it.

Jordan 46:36

Oh my gosh.

Alban 46:37

And he quit. And look, I feel for it, like that's not funny for the person who quit their job. But you've got to be somewhat grounded in reality. Like, go out and play some golf and go, I've watched PGA golf before and I've watched me play golf before. I know that these are not the same thing. And so if my podcast was coming back with, yeah, Kevin talks 80% of the time, Jordan's always going on and on about Amazon music and how great it is. I'd just be like, none of those sound right.

Jordan 47:06

Yeah.

Alban 47:06

Yeah. And so you want to be yourself grounded in reality enough that you would notice I am not a PGA professional golfer. And whatever it's told me about which episodes perform best, like I'm looking at my own stats too. And some of this might seem off. When it seems off, you're trying to direct it back towards reality.

Kevin 47:25

At no point in any of these conversations did it tell me to drop you two as co-hosts and go solo. But if it had, I would, I hope I would have recognized that that would be a terrible decision.

Jordan 47:35

You're like, you know what?

Alban 47:36

Kevin, you were absolutely right. Not only are you carrying these episodes, you'd be better off without the dead weight of your co-hosts.

Kevin 47:44

All right. So uh episode planning. Episode planning. Um, AI helped us identify that we had, you know, some topic categories, title patterns, episode structure, conversational dynamics. It has all this now as background information. Yeah. And so I thought that we could push into episode planning a little bit.

Jordan 47:59

This is where I'm excited about it.

Kevin 48:01

Yeah. Uh it's a little scary, but I started with based on uh my best performing episodes. Sorry, I should have said our best performing episodes. Over already?

Alban 48:11

I think this wasn't hypothetical. I think it did tell you to it's like a let's delve into these relationships more, Kevin. Let's get rid of these guys. Sorry, sorry, sorry.

Jordan 48:19

What topics can I cover by myself?

Kevin 48:23

Based on our best performing episodes, suggest some new episode ideas. Now, I honestly did not think that this prompt was going to work very well. And so uh like a backup prompt that I had in my mind was what topics have we not covered recently?

Jordan 48:36

Yeah, yeah.

Kevin 48:37

But I felt like that was more like, oh, take some of the old stuff and give me a newer version of it.

Jordan 48:42

Yeah.

Kevin 48:42

And and I really wanted to give it kind of free reign to feel like it could go off and find something that we've never covered before. So I started with the first one, which is just based on our best performing episodes. What are some new episode ideas? And AI it did a good job of suggesting a whole bunch of stuff. I didn't go ahead and list out all the ideas that it had, but like it's a it's like an infinite list of topics that we could cover. You know, it comes back with like 10 at first, and then it says, Do you want to go on more? Do you want to talk about doing a series, like a three-part series and how we would break it up? Again, I've done this in the past and I was not impressed with the output. Doing this as sort of a closing exercise after you've populated all of this valuable data, explained the data in different ways, told it how to break it up and chunk it and look at it, and then doing this at the end gave me much better results. And so that's why I put it here at the very end in the proper context and the proper with the proper data and the proper like contextualizing of that data. I was really happy with the results it was able to give me at this point in our conversation.

Jordan 49:42

Yeah.

Alban 49:43

Yeah, I did the same thing with Josh's podcast, Jordan's husband's podcast. I grabbed all of his recent episodes and I put it in with a prompt. I was just like, based on this, determine what this podcast is about, which ones perform well, and then recommend the episodes that he should cover and give me a bit of an outline for them. Jordan, what was Josh's feedback on those? Or what was your feedback on those?

Jordan 50:07

They're really good, actually. And what was really cool about the ones that you provided me, it wasn't just like, oh, here's the topics you should cover. It also provided like, here's some of the key points or the sequence of events that you need to make sure that you touch on. Here's why it matters to your listener base, and here's why it's such like a hot topic. Yeah. So those were really cool. And it was really funny because some of those were actually my list of upcoming episodes I've planned for them.

Alban 50:33

So you can either use AI or you can be married to Jordan and have her produce your episodes.

Jordan 50:38

Same thing. All right.

The Optimistic Case For AI

Alban 50:40

So, Kevin, you're supposed to be the big picture guy. So, what's the big picture here? What's the big lesson?

Kevin 50:45

Okay, that's a good challenge. I did think about it ahead of time. I was hoping that you would ask me that question, Alban. It's a very good question. And you touched on it earlier, and I didn't want to cut you off. But all the time, the only conversations that I hear in the big podcasting circles is how is AI gonna impact podcasting? And it's always like AI voices, it's always auto-generated shows. It's always AI is gonna take over ads, and shows are gonna become 50% ads because all these AI ads are popping in and all this stuff. It's all this kind of negative stuff. It just feels gross. I don't, it always feels like none of it's positive. Yeah. None of it's positive.

Alban 51:21

I've never heard it gone, oh, you know what I'd love to do? Not record the episode and have AI use my voice and make it sound like me.

Kevin 51:28

Right. And it almost feels like we're good at analogies on this show, so why not throw in a random one here? But it's almost like we have this brand new technology, and it is like mind-blowingly amazing. And we're immediately, as an industry anyway, looking at the worst possible outcomes or uses or side effects of this thing. And it's sort of like if you were on the field the day the Wright brothers like took flight for the first time, turning to your friend and just being like, Oh, great, can you imagine what's gonna happen to the environment now? Because all these jet airliners are gonna be flying all over the whole world. Airplane crashes are up 10,000 percent because we had zero before. You just saw something absolutely like mind-blowingly amazing. How is this happening? And you immediately take it to the worst place possible.

Jordan 52:16

Yeah.

Kevin 52:16

We don't want to be those types of people. We want to think about like this is amazing. This is amazing technology. How can we use it like to make ourselves better, to make the world a better place? And it might have some negative uses for sure, but like, let's not focus on those things. Let's not make those the story of the day. Let's make the positive side the amazing thing.

Jordan 52:33

Yeah.

Kevin 52:34

So, like in the sense of AI is going to ruin all creativity and the arts are dead, and yada, yada, yada. I think this is a great example of it's not replacing creativity, it's helping helping us be more creative. You know, it's pointing us in new directions. It's it's recognizing patterns, it's helping us find out what's working and what's not, what areas can we improve? How do we make the show better? How do we reach more people? How do we just change a few words in this episode with title so that more people who are looking for content like this might find it? Yes. And that's exciting and that's optimistic. And that got me in a totally different mindset when I think about how can we use AI or how can AI impact the world of podcasting? I think that's exciting.

Alban 53:16

I've been in podcasting long enough that people for years all we said was, man, if I just got signed by Spotify or I got signed by Amazon Music and I got enough money, then my show could work because I'd have a marketing budget, I'd have an editor, I'd have a team built around me, and I'd have all these resources. And when used well, you've actually got some really incredible resources. We got a lot for our show out of Kevin playing with this for a weekend.

Jordan 53:41

Yeah.

Alban 53:42

Uh we can get some really good topic ideas. You know, you might get 20 and two of them are good because you're the expert here. You're going to actually look at it. It's like having a research assistant who might go off and come up with 20 ideas and you go, we might do two of these episodes. So a lot of the things we wished for for years have come true. Yeah.

Jordan 54:03

Most podcasters are hobbyist podcasters. Like this isn't their day job. They don't have days and weeks to spend making sure that their podcast is being promoted and it's marketed and they've got all these campaigns running. And if you just take an hour, maybe three hours to really drill down into your content and do this kind of stuff, and it can have a really big uh impact on your show. We actually had a listener send in a case study from her own experience, uh, staff at Geopats Abroad. So she has this client that she grew primarily through optimization rather than promotion. It takes so much less time to do this. So before this show had relaunched, they'd been active for a year and the downloads were mostly like in the double digits, which is good. It's really good. But the growth was really inconsistent and they weren't growing. And so what she did was she went back through, she made clearer episode titles, stronger descriptions, and made sure that like the metadata was good. There was like wider platform distribution. If you have AI look into this stuff, it can identify these pain points for your podcast. And after she went through and updated the episode titles and all that stuff, the downloads increased month after month. And she said that the show actually grew more than 10 times from its original baseline after that.

Alban 55:24

Yeah, she sent in some screenshots. She did.

Jordan 55:26

Yeah.

Alban 55:27

It's the type of charts you're dreaming about. They're really small on the left and it's up into the right over time.

Jordan 55:34

Yeah, and it's really good. And she said that she's done the exact same thing for her podcast and it still receives 100 to 500 downloads per month, even though she hasn't published new episodes in years, just because she has gone through and optimized the catalog and it and it does make a big difference. So if you're on the fence about, I don't know if I have time for this, man, I really feel like just spending a few hours on this for your podcast could have huge payoff.

Kevin 55:58

I love it. Now, one of the things throughout this conversation that I've had with AI over a couple days is it's encouraged us to do a better job of landing the plane, coming home and drilling home the lessons from like every topic before we switch topics or the entire episode, if we just stuck on one topic the whole episode. So I'm going to try my best to do that now. And that is, I hope every podcaster who's listening to the show understands like these tools are available to you now. And you have a massive amount of data. Again, like we called it a treasure trove of data that's available to you. Whether you've just done a few podcast episodes or you've done hundreds, uh, there's still insights to be gleaned, there's patterns to be recognized. And there's areas for uh improvement, there's areas for you to recognize that you're already strong in to continue to lean into. But this doesn't have to take up a whole day. It doesn't have to take up a whole weekend. It can just be an hour. If you start by exporting your episode list, like I did, just copying and pasting that from your BuzzBrout dashboard. If you have transcripts, download those and put it in. Give it a few simple prompts to help connect all that data together and then ask some questions or ask it what questions should I ask? Or just based on the data, what do you see? It's going to be a fun start. And I think you're going to discover things that you never noticed about your show and ultimately hopefully be a better podcaster at the end of the day. So I think it's a worthwhile endeavor. I hope that uh some of the insights that we were able to discover and share with you today were helpful. And I don't want to wrap your episode for you, Jordan, but I feel like I should just say keep podcasting. We have fan mail always. Oh, wait, see, this is why I can't be the host. I mean, Jordan missed. The listeners are screaming. Where's our fan mail? Where's our stay in your lane? Well, I landed the plane prematurely. We were gonna taxi for a few minutes as we go through Fan Mailville.

Fan Mail

Jordan 57:34

Captain Solely over the Hudson. All right, so we do have some fan mail, and we are going to read in this episode. Uh, first up, we have Andrew from Fabulously Delicious, the French food podcast, says, Hi guys, great episode, and can't wait for the bigger episodes. Sounds like this could really help. My podcast is into its sixth season and over 250 episodes. Congratulations, Andrew. Every year I see double the growth, and now I get around 400 plus downloads an episode in the first seven days, which is great. The question I'd like to ask about what you're saying around SEO is something I always struggle with. Is it Google or the podcast platforms that we are targeting to get views or both? And for titles, tags, and descriptions, I can't work out where the SEO is to go. All of them perhaps, but then I struggle to make it sound natural. Yeah, I think that this is a very common issue that a lot of podcasters have.

Alban 58:29

The opportunity is in the podcast apps because that's where people are mostly searching to start listening to an episode. Most of the apps, your highest leverage points are the episode title, the podcast title, and the podcast author tag.

Jordan 58:46

Yeah.

Alban 58:46

And so you're not trying to cram 40 keywords into each of them, but the keywords that are applicable, let's make sure they're in there. So go look at your title and read it. Maybe tell the title of your podcast to somebody who's never heard of your podcast. Say, what is this about? And in your mind, I want you to imagine what are people searching when they want to find your podcast? And do those line up. So Jordan just did this for her husband's podcast, which is about UFOs, and she knew Alien Mysteries is something that people search that kind of fit his podcast. And so she added Alien Mysteries to the title, and it had an impact.

Jordan 59:26

Yeah. It now rings number one.

Alban 59:28

And you change something like your husband's name, you change to his name, but then like UFO investigator or something. You gave him a title.

Jordan 59:35

Yeah, I uh I put some stuff in the author tech too, which I think you're technically not supposed to, but it worked. So I don't know. All right, well, we're testing things out. Yeah, don't take me off, please.

Alban 59:49

Another thing I've noticed is this is as a listener and a podcast searcher. I'm getting really good results when I search um using Chat GPT for podcast episodes of people that I want to hear episodes from. So I'm like, here's a topic I'm interested in. I found some designer that I thought had some interesting ideas. And I said, has he ever done any podcast episodes? And it came back with some links and I was able to go listen to them. I wouldn't have found them through a podcast app based on the titles. But I think it's because people who are putting full transcripts online, those are actually showing up inside of ChatGPT. So I think it's gonna be really good. And I'm really glad that we rolled out all these transcript updates because it's going to help people show up in Google searches as well. I think those don't stress too much when you're trying to optimize it. That's gonna happen naturally if you're putting full transcripts in and titling these well.

Jordan 1:00:47

And Andrew, we actually have a deep dive episode on this very topic, and it's something like how to get discovered in Apple Podcasts or something like that. I will link to that episode in our show notes for this episode so you can listen to it. It's very good.

Kevin 1:01:01

Steph from Geopets Abroad wrote in and said, if the audio file episode title, audio file name, and Buzzsprout episode title are different, which one gets sent through to the RSS feed? I've always wondered, but I don't know how to test it without going down a deep rabbit hole. They might be different to you. You think they're different. The reality is they're not different. Whatever file name is what you upload doesn't matter anymore. We rename it once it gets sent through to BuzzSprout and we give it a unique identifier. If anybody downloads it, we actually rename it to the title that you give it within BuzzSprout. And the title you give it within BuzzSprout is what ends up in the RSS feed, not on the enclosure URL itself, but that's a technical detail that you don't need to worry yourself with because all the podcast apps are smart enough to recognize this is what you're saying the title of the episode is. So I don't really care what the file name is. So it's an interesting question, but it's not important for you to understand or because BuzzSprout handles all that technical stuff for you.

Jordan 1:01:53

That's interesting.

Alban 1:01:54

Tracy Ann from Inside Out Connections, a wellness podcast, wrote in. Hi guys, I love your podcast. It's been extremely helpful for a new podcaster, especially the episode about editing. This has been game changing for me. Before, I was afraid to take out the clunky, boring segments. But after listening to them myself and then understanding what has greater impact for hearing your episode, I can now get clear on what the important message is for my listeners. While a bit of chit chat is real, too much is boring. So my question to you, and we're going lightning round, how to gain more traction.

Jordan 1:02:27

I think that this episode and the last episode about titling episodes should help a lot with gaining traction, especially on your older episodes. But I don't know if Tracy Ann has a ton of episodes. I think she started like New Year's Day or something this year.

Kevin 1:02:41

Yeah, and go listen to our uh 25 podcast marketing tips, unique marketing tips. A lot of those I think would apply really well for a wellness podcast specifically.

Jordan 1:02:48

Yeah.

Alban 1:02:49

Where should I advertise my podcast beside the usual Instagram socials account to get more followers?

Kevin 1:02:54

I'd like to put advertisements for podcasts in between the games that you play in the back of Ubers. No, I'm kidding. I'm kidding. Um again, I would, all joking aside, I would point you to that same 25 unique marketing ideas podcast episode that we did. Yeah. A lot of that stuff is great. And it covers like crazy off-the-wall ideas like dropping uh coasters with your QR code in bars and stuff like that, and maybe taking people from a bar to the gym the next morning. I don't know. Like you'll come up with some cool, fun ways from listening to that episode.

Alban 1:03:20

Yeah, I also would say Overcast or Castbox, anywhere that allows you to put ads inside the podcast apps, those seem to have the best ROI.

Kevin 1:03:28

Yeah. And the challenge specifically for wellness podcasts and online advertising for them is that there's a lot of wellness podcasts. And I'm not saying that's not a reason that you shouldn't do it. It's just that you might find more success doing some like hyper-local stuff. You're probably connected with a bunch of uh fitness or health gurus around your area. And so spreading the word locally in those circles and then letting it grow from there might be more effective, at least initially, than trying to compete in the massive world of health and wellness podcasts that exist online. How do you get people to follow and review?

Alban 1:03:58

Even getting some of my friends to do it is a punishment.

Jordan 1:04:01

Okay. A lot of times your friends and perhaps even family members are not your target audience. I know that I had a podcast that has millions and millions and millions of downloads, and I can tell you for a fact that not a single one of my friends has listened to that podcast. I they know about it, they're aware, but they're not gonna do it. Really, the only way you can get them to follow your podcast is if you physically take their phone and follow it yourself on their phone.

Kevin 1:04:27

100% you should do that. And leave yourself a review and a random videos. Yes. Yeah.

Jordan 1:04:32

Yeah. But we're gonna do a quick cast next week about call to actions, and that will cover asking your listeners to please review your podcast and making sure that you're very convincing in your call to action to do that.

Alban 1:04:45

And last one on the on the lightning round from Tracy Ann, uh, how do you get advertising on the show? Uh, we did a really good episode, I think it's episode 203, how to choose the right monetization strategy for your podcast. Jordan wrote a whole blog post on this that was really good. And we did this whole episode about it. So we'll link that in the show notes as well.

Jordan 1:05:05

Yeah.

Kevin 1:05:06

100%. And for a wellness podcast, consider just turning on listener support, at least in the very beginning. Don't worry about getting ads as much as you say, like you're providing real value for people. You're helping them change their lives, become healthier, become uh, who knows, working their way out of depression, losing weight, whatever you're doing, I'm sure it's impactful in a positive way for people. And so a lot of people, even if you only have a few listeners, might be interested in just supporting you on a monthly basis. And so think about starting there. How can I provide enough value that people are willing to maybe return something for me so I can keep doing this for more people? And that's the listener support model versus the ad model.

Jordan 1:05:38

Absolutely.

Alban 1:05:40

So there's two ways people talk about marketing for products. One way is I get all this data and I analyze the data. And then there's people on the other end who just like they send one email and they said, How did you find out about us? And that one email, how did you find out about us? get some really interesting stories. And those stories help. them inform how they should be doing more marketing. And I think about this all the time in relation to podcasts. How do I find new shows? And so I've written these stories up before for myself of how I found my most recent favorite new podcast. And I want to do an episode about how you found your most recent new podcast you subscribe to and you're really listening to. How did you find out about it? Tell us as much detail as you want. And then what I want to do is kind of almost do some little case studies of how people are actually finding podcasts and going from no awareness all the way to I really love this show. So what did it look like for you? Think of the show that you most recently subscribed to and really liked and hit the button to text the show and tell us that story.

Kevin 1:06:51

Right. And then hopefully we can identify some patterns between these and start to identify like some new marketing funnels for podcasters.

Post Show: No Honor Among Thieves and Teens

Jordan 1:06:58

All right. Until next time thanks for listening and keep podcasting. Kevin, I have a bit of a parenting question for you.

Kevin 1:07:10

Okay, bring it in I'll put on my uh my family counselor hat. Yeah. Let's go.

Jordan 1:07:14

My my daughter is turning 14 in about a week and I've noticed that there's this new thing happening where she is taking my clothing. And it's not just like she's like occasionally slipping something on or gets like mixed up. It's like she is fully taking my clothes and folding them up and putting them in her dresser or hanging them in her side of the closet.

Alban 1:07:36

But your daughter's doing laundry? Oh yeah.

Jordan 1:07:38

She's doing laundry but she's also taking mine and then I couldn't find some pants for a while and I found them in her hamper. So she wore them to school. And do your daughters do this do they steal your wife's clothing?

Kevin 1:07:50

Oh I thought you were gonna say steal my clothing. Okay, different answer.

Jordan 1:07:52

Do they steal your clothing?

Kevin 1:07:53

Do you okay dad jeans are in but do your sons do this?

Jordan 1:07:57

Is this just like a teenager thing?

Kevin 1:07:59

Yes. So first of all, congratulations because just the fact and and and remind yourself of this every time you can't find your favorite piece of clothing to wear because your daughter has stolen it. Remind yourself that this is a massive compliment that you have a very cool hip teenager and for whatever reason they like your style enough to take your stuff. Like that's a massive compliment.

Jordan 1:08:20

That's true.

Kevin 1:08:20

And so congratulations on being a cool hip mom and maintaining your sense of style. And number two, I also think that as frustrating as it can seem at times, I think we have to fight the urge to sort of lash out about it because it's like this cool bonding thing. So I do have I have two sons. One of them will take my clothing a lot more than the other but they'll both take stuff from time to time. And I feel like I don't know it's just like another connection point. As your kids get older and older your job is to help prepare them to go off and be independent of you. Right? Your job is to prepare them for the world. And it's the hardest thing in the world because they are like it's like literally cutting off or like pieces of your body breaking off from you. That's the pain that we experience as parents as we're pushing these children away and preparing them for the world. And the fact that they want to take something that I own that I wear on my body and put it on their body and then go out into the world wearing that thing is frustrating because I'm like that's I don't know we're going to play golf and it's like where's the golf shirt I want to wear today and then my wife is like oh Evan wore that to play golf yesterday.

Jordan 1:09:22

Or they're wearing Banana Republic to middle school.

Alban 1:09:24

And you're like Kevin's like where are my two double pop collar golf shirts oh my son's wearing them sometimes I think do I want this problem or the exact opposite problem and the exact opposite problem is you have such bad taste in clothing your kids never would take it and they make fun of it and they say oh my gosh that's so embarrassing. You've actually got the better of the two problems the one that hey that actually looks pretty sweet mom. That looks so good that ethically I'm willing to bend some uh ethical rules and steal.

Jordan 1:10:01

Well you know what I actually found a little bit of a benefit to this because last time we went shopping she was trying to talk me into this really cute jacket and I was like no no I'm not spending that and she's like mom we both could wear it and I was like wait a second we both could wear this so your daughter is 14 now yeah right she's turning 14 and she's taller than me.

Kevin 1:10:21

So let me give you some insight into the next phase.

Jordan 1:10:23

Okay.

Kevin 1:10:24

So right after the discovery that you just made the next thing that happens this just happened Evan and I were in a store together and he you know works a little bit now he's got a little bit of money and we both saw uh again it was this golf shirt that we both liked and it was like expensive and I'm like I I like the shirt but I don't want to pay that for it and Evan's like the same thing. And I was like but we could split it and he's like well you're like a lot older and have a real job so maybe you should pay like 80% and I pay 20%. So we had a little negotiation I think we named it what's he talking about we landed somewhere around 60 40 so you know it's like a hundred dollar shirt he put in 40 bucks I put in 60 bucks and we both have a shirt now. Oh man parenting hacks are amazing yep so I think it's a compliment Jordan thanks for sharing it's a fun phase it's fun