ChatGPT Prompts for Podcast Show Notes: 25 Templates

The short version: good podcast show notes come from feeding ChatGPT your actual transcript and specific instructions, not asking it to "write show notes" from a two-line summary. Below are 25 prompt templates covering summaries, timestamps, titles, quotes, guest bios, and repurposing, plus the five-step workflow I use to get a full set of notes done in about 12 minutes instead of 45.

Why I stopped writing show notes from memory

Between 2021 and 2023 I recorded 92 episodes of my own podcast. For the first 30 or so, I wrote show notes the way most hosts do: listened back at 1.5x speed with a notepad, tried to remember the best line the guest said, guessed at a title, and published something serviceable about four hours after the recording ended. It worked. It also took forever.

I've also been a guest on more than 300 podcasts over the years, which means I've seen show notes from the other side, as the person whose 45-minute conversation gets turned into three sentences and a Spotify link. Most of those notes are thin. Not because the host is lazy, but because writing a decent summary of an hour-long conversation you just finished recording is tedious work, and tedious work gets rushed.

ChatGPT fixed the tedium, not the thinking. That distinction matters more than most guides on this topic admit.

The uncomfortable bit nobody selling you a "10x your show notes" course wants to say

Here's the truth: hardly anyone reads full show notes. Download numbers come from the title, the artwork, and the host's existing audience, not from a beautifully formatted summary on your website. I've checked analytics on my own episode pages. Time on page for a 1,200-word show notes post averages under 40 seconds. People skim the first two lines, glance at the timestamps to find the bit they care about, and leave.

So why bother with 25 prompts for something so few people read closely? Because show notes aren't really for the listener scrolling on the day. They're for Google, for the guest who wants a quote to share, for the sales team who needs a one-line summary six months later, and for you when you're building a "best episodes" page and can't remember what happened in episode 41. Show notes are infrastructure, not content. Once you treat them that way, the prompts below make a lot more sense, and you stop trying to make ChatGPT write something "engaging" when what you need is something accurate and searchable.

The five-step workflow (do this before you touch a single prompt)

None of the 25 templates below work well if you skip this part.

• Step 1: Export the raw transcript. Descript, Otter.ai, or your host's built-in transcription (Riverside and Buzzsprout both do this now) all produce something usable in under five minutes.
• Step 2: Clean it lightly. You don't need perfect punctuation, but strip out "um," repeated words, and cross-talk if the tool didn't already. Two minutes.
• Step 3: Paste the full transcript into ChatGPT first, with no instructions, and just say "Here is a podcast transcript. Confirm you've read it." Then send your prompt in the next message. This stops it from summarising the first 500 words and guessing the rest.
• Step 4: Run three or four of the prompts below against the same transcript in the same chat, so it keeps context.
• Step 5: Edit by hand for 5 to 10 minutes. Add the one detail ChatGPT can't know, like the guest's actual company name if it misspelled it, or the fact that the funniest moment came right after a joke about your dog.

That whole process takes me about 12 minutes now, down from the 45 I used to spend writing notes solo. If you want more prompts built specifically for the transcript-to-notes pipeline, I go deeper on that exact process in ChatGPT prompts for podcast show notes that people click, from the transcript up.

Summary and overview prompts (1 to 3)

• 1. The three-sentence hook: "Using this transcript, write a 3-sentence summary that a stranger could read to decide if they want to listen. No jargon, no 'in this episode we discuss.' Start with the most surprising claim the guest made."
• 2. The 150-word overview: "Write a 150-word show notes overview from this transcript, written in a warm, first-person tone as if I'm telling a friend about the conversation. Avoid the phrase 'dive into.'"
• 3. The problem-solution frame: "Summarise this transcript by identifying the specific problem the guest solves for their audience, and the one method or framework they described for solving it. Keep it under 100 words."

Timestamp and chapter prompts (4 to 6)

• 4. Basic timestamps: "Go through this transcript and create timestamped chapter markers every time the topic shifts. Format as [00:00] Topic name. Aim for 8 to 12 markers for a 45-minute episode."
• 5. The "find the gold" version: "Read this transcript and flag the three moments most likely to make someone stop scrolling and start listening. Give me an approximate timestamp and a one-line description for each."
• 6. Question-based chapters: "Turn this transcript into chapter markers phrased as questions the listener will get answered, e.g. '[12:30] How do you price a first consulting call?' Pull the actual questions I asked as the host."

Title and SEO prompts (7 to 9)

• 7. The search-intent title: "Suggest 8 episode titles for this transcript that someone might type into Google or YouTube search, not just a clever pun. Include at least 3 that contain a number or a specific outcome."
• 8. The meta description: "Write a 155-character meta description for this episode's show notes page, written for someone who has never heard of the guest, that would make them click from a Google search result."
• 9. Keyword extraction: "From this transcript, list the 10 phrases or topics a listener searching Google would use to find this exact conversation. Prioritise specific terms over general ones."

Guest bio and credibility prompts (10 to 11)

• 10. The two-line bio: "Using only what the guest says about themselves in this transcript, write a two-sentence bio suitable for show notes. Don't invent job titles or company names that weren't mentioned."
• 11. The credibility pull: "Find every specific credential, result, or number the guest mentions about their own work in this transcript (years of experience, clients served, revenue figures, awards) and list them as bullet points."

Pull-quote and highlight prompts (12 to 14)

• 12. Quotable lines: "Pull 5 direct quotes from the guest in this transcript that would work as standalone social media graphics. Each should be under 20 words and shouldn't need context to make sense."
• 13. The contrarian quote: "Find any moment in this transcript where the guest disagrees with common advice in their industry. Quote it directly and note the approximate timestamp."
• 14. The vulnerable moment: "Identify one moment in this transcript where the guest shares a mistake, failure, or something they'd do differently. Quote it word for word."

Full show notes body prompts (15 to 17)

• 15. The structured long-form: "Write full show notes for this transcript with these sections: Overview (100 words), What You'll Learn (5 bullets), Key Moments (3 to 5 with timestamps), About the Guest (2 sentences), Resources Mentioned (bulleted list of anything named in the conversation)."
• 16. The narrative version: "Write show notes as a short narrative, 200 words, describing how the conversation unfolded, written in first person as the host. Don't summarise every point, just the arc of the conversation."
• 17. The skeptical listener version: "Write show notes aimed at someone who thinks podcasts are mostly fluff. Focus only on the specific, actionable takeaway from this transcript, stated plainly in one paragraph."

Resources and CTA prompts (18 to 19)

• 18. Resource extraction: "List every book, tool, website, or person mentioned by name in this transcript, in the order they appear, with a brief note on the context in which each was mentioned."
• 19. The next-step CTA: "Based on this transcript, write one sentence inviting the listener to take a specific next action (follow the guest, try the method discussed, listen to a related episode), avoiding generic phrases like 'don't forget to subscribe.'"

Repurposing prompts (20 to 23)

• 20. LinkedIn caption: "Write a LinkedIn post, 100 to 150 words, based on the single most useful idea in this transcript, written in my voice: direct, first person, no corporate tone, ending with a question for engagement."
• 21. Newsletter blurb: "Write a 60-word blurb about this episode for an email newsletter, written to make a busy subscriber click through even if they only have 5 seconds to decide."
• 22. YouTube description: "Write a YouTube video description for this episode, 300 words, front-loading the most important keywords in the first two sentences, followed by timestamps and a short guest bio."
• 23. Twitter/X thread: "Turn the three best insights from this transcript into a 5-tweet thread, with the first tweet as a hook that doesn't reveal the full insight."

Editing and tightening prompts (24 to 25)

• 24. The cut-by-30% edit: "Take this draft of show notes and cut the word count by 30% without losing any specific numbers, names, or quotes. Cut adjectives and filler phrases first."
• 25. The AI-tell remover: "Rewrite this draft to remove phrases that sound like AI wrote them (delve, unlock, game-changer, in today's fast-paced world) and replace them with plainer language."

Where this saves you money, not just time

If you're running a podcast as part of a bigger content or lead-generation strategy, show notes are one small piece. The bigger question is usually how AI fits into your marketing across the board, which is worth thinking through rather than bolting on prompt after prompt. I've written a fuller breakdown in how to use ChatGPT for marketing in 2026 if you want the wider picture. And if you're at the point where you're weighing up whether to keep doing this yourself or bring in help to set up the whole system, that's exactly the conversation I have with clients through my AI consultant for small business work.

One more practical note: these prompts assume you're the one prompting, but if someone on your team is new to ChatGPT entirely, don't hand them 25 templates and walk away. Start them on something simpler first, like the basics covered in 40 simple ChatGPT prompts for beginners, then layer these on once they're comfortable pasting a transcript in without panicking about formatting.

What I got wrong for two years

For the first year I used ChatGPT for show notes, I made the same mistake almost everyone makes: I asked it to write show notes without giving it the transcript, just a rough two or three sentence summary I typed from memory. The output was fine, grammatically. It was also generic, because ChatGPT was inventing detail to fill gaps, and inventing detail in someone else's podcast episode is a fast way to publish something wrong. The guest's company got renamed. A statistic got softened into a vague "many people." Once I started pasting the full transcript every single time, the accuracy problem disappeared, and the notes sounded like the conversation that happened rather than a conversation that could have happened.

The second thing I got wrong: treating every prompt as a one-shot. The best notes come from running 4 or 5 of the prompts above in sequence, in the same chat, so ChatGPT has full context by the time you ask for the LinkedIn caption or the newsletter blurb. Starting a fresh chat for each piece wastes the context you've already built.

A quick real example

On episode 61 of my old show, my guest was a former agency owner who'd shut down her business after a bad year and rebuilt as a solo consultant. The raw transcript ran to about 9,000 words. Using prompt 13 (the contrarian quote prompt), ChatGPT pulled this line almost immediately: "Everyone told me to hire faster to grow. Hiring faster is what nearly killed the business." That one sentence became the pull-quote graphic, the newsletter subject line, and the opening line of the show notes. Without a targeted prompt pointed at a full transcript, I'd have had to scrub through 50 minutes of audio to find it myself, or more likely, I'd have missed it entirely and led with something safer and less interesting.

If you want a broader set of role-specific prompts beyond podcasting, the general collection in 60 ChatGPT prompts across 12 professions is worth a look, and if your bottleneck isn't show notes but your whole week feeling unmanageable, 23 ChatGPT prompts for productivity covers the scheduling and planning side I use around recording days.

Related reading: It's Not About the Money.

This guide is part of my ChatGPT Prompts and Limits: 30 Guides to Fun Prompts, Photo Edits and Daily Caps. To write your own, try the free ChatGPT prompt generator.

Related: writing for us on podcasting.

Related: the no code page.

Related: working with an should you hire an ai consultant what they do.

Frequently asked questions

How long should podcast show notes be?

Somewhere between 150 and 400 words for most episodes, plus timestamps. Longer notes help SEO if they're specific and full of real names, numbers, and quotes from the episode, but padding a summary out to 800 words with generic sentences won't help rankings or readers.

Can ChatGPT write show notes from just an episode title, without a transcript?

It can, but it will invent details to fill the gaps, which means factual errors about your guest's company, numbers, or claims. Always paste the actual transcript first and generate notes from that, even if it takes an extra five minutes to export it.

What's the fastest way to get timestamps without listening to the whole episode again?

Export the transcript with timecodes included (Descript and Riverside both do this automatically), then use a prompt like template 4 above, asking ChatGPT to mark chapter shifts using the existing timecodes rather than estimating them itself.

Do podcast show notes help with SEO?

Yes, but mainly for long-tail search terms tied to your guest's name, specific advice given, or niche topics discussed, not for competitive keywords. A well-written notes page with real quotes and specific numbers can rank for searches your episode audio never would.

Further reading

• OpenAI's own platform documentation