TL;DR — I built a Claude skill that takes your podcast transcript and creates very good drafts of a podcast trailer. It doesn’t do the actual editing of course, but it gives you multiple versions of a script and pulls a lot of usable lines with timestamps. It saves me tonnes of time.

Get the skill for free here below

(it also works in Chat GPT, Gemini and other LLMs)

The most time consuming part of podcast post-production process is building a captivating trailer that hooks viewers and keeps them watching.

I used to spend hours marking down timestamps, taking notes on the side, clipping soundbites and searching through transcripts trying to create a coherent structure that would pull the viewer in, confirm that this video is where they want to be, set up the narrative for the episode, and leave them with a captivating cliffhanger.

Now Claude does the difficult work for me, I’ll show you how.

First of all, you don’t need a trailer for your podcast.

Many podcasters seem to believe it is a necessity to have a flashy intro-trailer that runs before their podcast. They became popular following the success of Diary of a CEO, which is evident by the fact that when you search ‘podcast trailer’ on Google the top results all reference DOAC.

Just because they do it, you don’t need to do it.

If you can start your podcast in a way that follows the four principles I’m going to talk about here, then you don’t need one. Even with the help of AI, they still take a long time and the ROI is not there for most people.

But if you still want an intro-trailer for your podcast, here’s what it should accomplish:

Confirm the click: Within the first few seconds, prove the video is what the title and thumbnail promised, so the viewer knows they're in the right place and haven’t been clickbaited. If the title is ‘How to Make $100k with Instagram Brand Deals’ make sure the opening line is related to brand deals, or a credible lines about large revenue sums.

Earn trust: Show why this guest is worth listening to through results, numbers, accolades and genuine host reactions, do not give a biography. We just need the audience to understand that this person is a credible voice of authority on this particular subject.

Demonstrate value: Give a taste of a specific tactic or success story, so the viewer knows the episode gets into real detail. The trust is beginning to form, we’re going to build on that by showing that they’re going to get value from this episode.

Open a loop: End on an open loop, such as a question or a cut-off sentence, that only the full episode can close… a cliffhanger.

Trailer lengths can vary. They can 30 seconds, they can 3 minutes. Do what feels right to you, test different lengths and watch your retention curves. I usually aim for 30-60s.

If you think you can open your podcast in a way that hits these key points, without a fancy trailer, do that… it’s save you a lot of time and possibly money

The Claude skill I’ve shared here is essentially built on these four principles but with a lot more context.

I’ve edited over 100 podcast transcripts alongside Claude, refining my instructions and prompts to create a system that works for me, and hopefully others too.

All you need to do is

It’ll take a few minutes to read your transcript and execute the skill. Once complete you’ll now have a .MD file, which can be viewed within Claude or opened in another application such as Google Docs or Notion.

I prefer to open the document in Google Docs as it’s easiest to navigate. The document will be many pages long, so switch to pageless mode and use the document tabs.

The document that Claude provides you will find the following:

There will be a lot of them. For an hour long episode you may have over 50 possible options for each segment. Many of them can be disregarded and often the top suggestions are the best.

Here’s what that looks like. This is the list of hook suggestions. Straight away the first options are suggestion we open with brand partnerships and revenue streams etc. A good hook since the episode packaging is likely to be framed around that topic.

Claude will then provide some useful extras:

Finally, at the bottom of the document you will find some pre-assembled trailer suggestions. You will find:

Most of the times that you use this skill, you will probably just be able to scroll to the bottom, find the long script and cut what you like from there. It does a really good job of selecting the best lines and leaving the rest behind.

Now your job is to locate these soundbites in your timeline, pull them out and start constructing your trailer.

There’s no way to inject the human element into a Claude skill. It does not know the subtleties of human conversation, it does not know your audience, it cannot understand pacing and emotion. So you still need to edit yourself. Sorry.

But the skill cuts out a lot of time and gets you a very very good first draft.

For approximately 80% of the trailers I edit, the final version is some variation of what Claude suggested. Of course I move segments around, cut parts it suggested, sometimes changing the whole structure. But I didn’t have to spend hours searching through the episode for those ‘best bits’.

You’re also back on your own after this point.

Sound design, visuals, animations. I can’t help you there.

A lot of that stuff isn’t necessary anyway, it’s flair. It may tell the audience ‘this is a high-production show’, but the message is the same whether there’s music and animations or not.

If you would like to see the finished product of a podcast trailer used with the assistance of this Claude skill, this is a recent episode that I edited.

If you try the skill and have any feedback, let me know. I’m interested to hear how it helps you!