TL;DR: To promote a new podcast episode on social media, don’t post “New episode out now” with a link. Pull three things from the recording (a short clip, one quotable line, one question it answers), turn each into its own post for the network it suits, and spread them over the first week. An AI agent can draft and queue them from your transcript, but you choose the clip.
The episode went live Tuesday morning. By lunch you had pasted the Apple Podcasts link into X, LinkedIn and a Discord server, each time with the same sentence, and the total effect was two likes and a friend saying “nice.” The recording took three hours, the editing took four, and the promotion took ten minutes of copy-paste that nobody noticed.
This is a guide for people who release an episode every week or two and promote it alone. It covers what to extract from the recording, how each network treats the post, and which parts you can hand to an AI agent without it sounding like one.
Why “new episode out now” posts get ignored
A feed is full of people deciding in about a second whether to stop. A title and a link gives them nothing to stop for, because the title is a topic and the point of the episode is somewhere in minute 31.
Most promotion guides make the same point in different words: reshare each episode several times and in several shapes rather than once at launch. Podcast Episode Promotion Deserves More Than One Week argues for exactly that, and Buffer’s podcast promotion guide covers the wider list of tactics. Neither gives you a number you can treat as law, and I’d be wary of any article that quotes “audiograms get 5x the traffic” without a study behind it. What you can rely on is the logic: a post that carries one idea gets read, and one that carries a link gets scrolled past.
The three posts: clip, quote, question
Here’s the routine that holds up across shows. Pull three assets from each episode:
- A clip. Twenty to forty seconds where the guest says something specific, a story or a disagreement rather than a greeting. Audiogram or captioned video, whichever your editor exports fastest.
- A quote. One sentence, as text, that makes sense with zero context.
- A question. The question the episode answers, phrased the way a listener would type it.
Each one becomes a different post. The clip goes where video plays natively, the quote goes where text is the format, and the question goes wherever people ask questions. You are not writing one announcement four times, you’re writing three different posts.
What each network does with a podcast post
| Network | Best-fit post | Notes |
|---|---|---|
| The quote plus 3-5 lines on what you learned or what the guest got wrong | Story first, link near the end | |
| X | The quote, or the clip if you have it | Link in the post is normal; expect it to travel less than a native clip |
| Bluesky / Mastodon | The question plus the link | Link cards show a preview; both audiences tolerate links |
| Nothing, unless a thread asks the question your episode answers | Answer in the comment first, mention the episode second | |
| Threads | The quote | Short text suits the format |
| Newsletter / Discord / Telegram | A personal note to people who already follow you | The link is the point here |
Reddit rewards answers, not announcements
Reddit has no site-wide limit on self-promotion, but its informal guide, Reddiquette, suggests that only a small share of what you post should be your own content, and many subreddits ban promotion outright. The version that works is boring: someone asks a question your guest answered, you give a real answer in two paragraphs, and you add “we went deeper on this in an episode” at the bottom. Our Reddit automation guide covers the tooling, but whether to post is your call, not the scheduler’s.
LinkedIn wants the story behind the episode
A LinkedIn post that works for a podcast reads like a note to a colleague: one thing the guest said that changed your mind, one line of context, the link. For scheduling details see scheduling LinkedIn posts.
Spreading it over the first week
Here is a schedule that doesn’t take over your week:
- Release day: the LinkedIn story post and the X quote. Reply to comments yourself.
- Day 2: the clip, on whichever network your audience watches video.
- Day 4: the question post on Bluesky and Mastodon.
- Day 7: a different quote from the same episode.
The point of staggering is that each post gets its own reaction window, and a follower who sees three in a week sees three different things. The queue mechanics are in batch scheduling social posts.
Where an AI agent fits
Extracting the three assets is a judgment call. The clip is the one you’d play to a friend, and an AI can’t hear that for you. But once you’ve marked the timestamp and pasted the transcript, the rest is mechanical: writing four versions of the quote at the right lengths, formatting them per network, and posting on a schedule.
PublishPort is built for that handoff. You sign in to your own accounts in the desktop client, and your AI agent calls list_capabilities and local_bash through the relay, reading each command’s --help to learn the syntax, then posts from your own logged-in sessions. You give it the transcript, the clip timestamp and the links, and it returns drafts to approve and queues the approved ones. Plans are on the pricing page. The same flow works for other recurring announcements: video launches, GitHub releases and RSS feeds.
Limits to plan around:
- The desktop client has to be running and logged in when a post is due.
- Text and link posts are the dependable path. Attaching a rendered audiogram video to each network is a separate job, and we don’t claim it works everywhere.
- Models paraphrase confidently and sometimes wrongly. Check every quote against the audio, since a misquoted guest is a real problem.
- Run a dry-run on a throwaway post first; testing an AI agent before it posts lays out the routine.
Episode promotion checklist
- Clip, quote and question picked from the recording
- Every quote checked against the audio and approved by you
- Each network gets a different post, not the same text
- Reddit only where a real question exists
- Posts spread over seven days
- Desktop client running and logged in
- Time blocked to reply to comments
FAQ
How many times should I promote each podcast episode?
More than once, and not all on release day. Guides such as this one on promotion beyond the first week recommend repeating episodes in different shapes. Three to four posts across the first week, each with a different angle, is a sensible starting point to test against your own numbers.
Do I need audiograms to promote a podcast?
No. A quote post costs nothing and works on every network. Clips help where video plays natively, but a weak clip does worse than a good text post, so pick the clip carefully or skip it.
Should I post my podcast link on Reddit?
Only where it answers a question people are asking and the subreddit’s rules allow it. Answer the question in the comment first, mention the episode second, and say it’s yours.
Can I schedule podcast promotion before the episode is live?
Yes, as long as the post time falls after the episode is published and you’ve checked the link resolves. Directories like Apple Podcasts can lag behind your host’s own page, so link to the host page first.
Can an AI write my podcast show notes and social posts?
It can draft them well if you feed it the transcript. Without that it writes generic copy about the topic. Read each draft against the recording before anything is posted.
Is it worth promoting old episodes?
Often, yes. A quote from an older episode that fits something currently in the news is cheap to post, and most of your followers haven’t heard the back catalogue.