Ship a podcast from idea to RSS feed
An 11-step playbook. Open in Dock and you'll get four surfaces:
- **Steps** (table) — 11 gates from premise to live RSS feed in directories
- **Episode pipeline** (table) — one row per planned episode: title, guest, status, ship date
- **Brief** (doc) — your show premise, format, voice, and the audience profile
- **Asset library** (table) — cover art versions, intro / outro music, sound effects, episode trailers
Read `Steps` top-to-bottom. The first 3 steps decide whether the podcast survives episode 12.
Outcome
Your podcast live in Apple Podcasts + Spotify with 3 episodes published, an RSS feed under your control, a sustainable production rhythm, and the agent-assisted show notes that drive shares.
Estimated time: 3-4 weeks setup + ongoing 4-8 hours per episode
Difficulty: intermediate
For: Founders + creators launching their first podcast.
What you'll need
Pre-register or install before you start.
- Riverside ($15/mo Standard, free tier with 2hr/mo) — Browser-based recording with separate-track local capture (better than Zoom for podcast audio).
- Descript ($15/mo Creator) — Edit audio by editing the transcript; AI-powered cleanups (filler words, gaps).
- Transistor ($19/mo Starter (unlimited episodes)) — Podcast hosting + RSS + analytics + automatic submission to directories.
- Buzzsprout (Free tier (2 hours/mo, 90-day retention) or $12/mo for 3 hours) — Alternative podcast host with strong free tier and beginner-friendly UI.
- Shure MV7 ($249) — USB + XLR mic, the most-recommended starter pod mic for solo + remote shows.
- Auphonic (Free tier (2 hr/mo) or $11/mo for 9 hr/mo) — Auto-leveling, noise reduction, loudness normalization for finished episodes.
- Canva ($15/mo Pro) — Cover art + episode-specific graphics for social.
The template · 11 steps
Step 1: Define the premise that will sustain 50 episodes
Estimated time: 1-2 days
The single most-skipped step is also the most-important. Most podcasts die at episode 12 because the premise was 'a podcast about [broad topic]' and the host ran out of fresh angles. A premise that sustains 50 episodes is narrow enough to repeat without staleness ('first 90 days as a CTO', 'how solo founders shipped their first $10k month') and broad enough to fit 50 different stories. Write the premise in one sentence; if it doesn't pass the 'could I do 50 episodes' test, it's too narrow or too broad.
Tasks
- Draft 5 candidate premises in one sentence each
- For each, list 10 hypothetical episode titles and 10 ideal guests; cut any premise that hits the wall
- Test each premise by sending it to 5 friends and asking 'would you listen weekly?'
- Pick the one that survives + has the most clear-cut episode flow
- Document the premise + the next 10 episode candidates in the Brief doc
[!CAUTION] Gotchas
- Premises like 'a podcast about marketing' fail by episode 12. Specificity wins; 'how indie SaaS founders found their first 100 customers' is a sustainable premise.
- If the premise depends on you having an A-list rolodex, the early episodes will look thin. Pick a premise that works with B-list guests too.
- Premises that overlap with established big-podcasts ('how I built this' competitors) need a sharper angle, not just more episodes.
Step 2: Pick the format and the cadence
Estimated time: 1 day
Format: solo monologue, two-person conversation, interview, panel, narrative-edit, AMA. Each has very different production cost. Solo monologues are 1-2 hours of work per episode; high-edit narrative shows are 10-20 hours. Cadence: weekly is the discoverability sweet spot; bi-weekly is sustainable for solo creators; daily is for the burnout brigade. Pick the format AND cadence that you can sustain for 12 months.
Tasks
- Pick the format: solo / interview / panel / narrative
- Estimate the production hours per episode for that format (record + edit + show-notes + ship)
- Pick the cadence based on hours-per-episode and your weekly bandwidth (12 hrs/week available = bi-weekly works for interviews, weekly works for solo)
- Decide on episode length: 20-30 min is the sweet spot for new podcasts (lower commitment from listeners)
- Document the format + cadence + length target in the Brief
[!CAUTION] Gotchas
- Daily podcasts have higher discoverability but you'll burn out by month 4 unless this is a full-time job.
- 60+ minute episodes have 30-50% lower completion rates. New listeners drop off; loyal listeners stay. New podcasts need new listeners.
- Format-switching mid-show (solo to interview, etc.) confuses subscribers. Pick one and stick with it for the first 25 episodes.
Step 3: Buy the gear (don't overspend)
Estimated time: 1-2 days
The single biggest gear-money mistake is buying a $2000 setup before episode 1. The starter kit that sounds great: a $250 mic (Shure MV7 or Samson Q2U), $20 of acoustic foam for your recording space, $30 closed-back headphones, and a quiet room. That's it. Spend the saved money on editing software and the cover art instead.
Tasks
- Buy a USB or USB+XLR mic: Shure MV7 ($249) or Samson Q2U ($69)
- Buy closed-back headphones: Sony MDR-7506 ($99) or Audio-Technica ATH-M40x ($99)
- Treat your recording space: 4-6 panels of acoustic foam ($20-50) on the walls behind you
- Test the room with a clap-test (no echo, no buzz) — if it echoes, add foam
- Verify the recording chain: mic → audio interface (or USB direct) → recording software (Riverside or QuickTime)
Pointers
- [Official] Shure MV7 specs
- [Official] Samson Q2U (budget pick)
[!CAUTION] Gotchas
- A $2000 mic in an untreated room sounds worse than a $69 mic in a treated room. Spend on the room first.
- AirPods for podcast recording sound like phone calls. Even a $69 USB mic dramatically beats them.
- USB mics that connect via USB hub get electrical interference. Plug directly into the laptop or use a powered hub.
Step 4: Design the cover art (the most-underrated growth lever)
Estimated time: 1-3 days
Cover art is the thumbnail listeners see in directory grids. Bad cover art means no clicks; clicks compound to subscribers. The constraints: 3000x3000 px, JPEG or PNG, RGB, must read at 100x100 (the smallest size in directories). The pattern that works: high-contrast background, large legible title, simple iconography, strong brand color. Test the art at 100x100 before submitting.
Tasks
- Design 3-5 cover art candidates at 3000x3000 px
- Test each at 100x100: does the title read? Does the iconography survive?
- Iterate based on which survives the small-size test
- Keep the title large (occupies 40-50% of the artwork)
- Avoid busy backgrounds, drop-shadow text, and complex illustrations — they fail at 100x100
Pointers
- [Official] Apple Podcasts artwork specs
[!CAUTION] Gotchas
- Cover art with white backgrounds disappears in directory grids. High-contrast color blocks dominate.
- Photos of the host(s) face often outperform abstract designs for early podcasts (faces drive more clicks).
- Apple Podcasts rejects artwork with text in URLs, episode numbers in the artwork, or low-resolution images. Read the spec.
Agent prompt for this step
Critique 3-5 podcast cover art candidates.
Inputs (the user attaches images):
- 3-5 candidate cover artworks at 3000x3000 px
- The show premise + audience from the Brief
For each candidate:
1. Does the title read at 100x100 px? (Mockup mentally or describe what would be lost.)
2. Does the iconography survive the small-size test?
3. Color contrast: high or low? Saturation: balanced or muddy?
4. Brand consistency: does it match the show voice (irreverent / serious / technical)?
5. Pattern-match: which existing top-100 podcasts does it visually echo? Is that intentional?
Output:
- A ranking of the candidates from strongest to weakest
- 3 specific iteration suggestions per candidate
- A recommendation on the top 1-2 to ship
Constraints: be specific. 'Too busy' is unhelpful; 'the title competes with the photo background, increase title size by 30% and dim the photo' is actionable.
Step 5: Write and record the trailer (episode 0)
Estimated time: 1 day
Before episode 1, ship a 60-90 second trailer that introduces the show: who you are, what the premise is, who it's for, who's coming up. Trailers serve two purposes: (1) submit to directories with a trailer + 1 episode meets the minimum threshold, (2) listeners who land on the show page hear the trailer first and decide to subscribe. Trailers underperform full episodes for downloads but compound subscribers.
Tasks
- Write the trailer script (1 page, 60-90 sec at speaking pace)
- Structure: who-I-am (15 sec) + premise (20 sec) + who-it's-for (15 sec) + first-3-episodes-tease (20 sec) + CTA (5 sec)
- Record + edit the trailer to a tight 60-90 sec
- Add intro music + outro music (royalty-free from Epidemic Sound, Soundstripe, or your own)
- Save in the Asset library as the canonical trailer
[!CAUTION] Gotchas
- Trailers > 90 sec lose listeners halfway through. Cut to 60-90 max.
- Don't record the trailer first chronologically; record episode 1 first so the trailer can preview the actual show voice.
- Trailers that just say 'subscribe' without telling listeners what they'll get underperform versus 'here are the first 3 episodes you'll hear'.
Step 6: Record episodes 1-3 before publishing any
Estimated time: 1-2 weeks
Banking 3 episodes before launch lets you ship weekly without the production stress catching you. The first 3 episodes also let you debug the format: episode 1 will be the worst, episode 2 will be 30% better, episode 3 will be the version that defines the show. Don't ship until you have all 3.
Tasks
- Plan + record episode 1 (your strongest hook story or guest)
- Plan + record episode 2 (the format-debug episode)
- Plan + record episode 3 (the polished version that becomes the template)
- Edit each in Descript or Logic Pro: cut filler, level audio, normalize loudness to -16 LUFS
- Run each through Auphonic for final loudness + noise reduction pass
Pointers
- [Tool] Descript editing
- [Tool] Auphonic loudness normalization
[!CAUTION] Gotchas
- Episode 1 will sound rough. Don't ship it as-is; re-record after episodes 2-3 are done so episode 1 reflects your dialled-in format.
- Long unedited pauses + filler ('um', 'so', 'you know') drop listener completion rates. Edit aggressively.
- Loudness normalization to -16 LUFS is the podcast standard. Episodes outside this range sound 'wrong' on phone speakers.
Step 7: Pick the host and configure the RSS feed
Estimated time: 2-3 hours
The RSS feed is the canonical source of your podcast. Hosts (Transistor, Buzzsprout, Anchor) generate the RSS, host the audio, submit to directories, and provide analytics. Pick one with: unlimited episodes, your own custom domain support, no platform-locked content. Anchor (Spotify-owned) has limitations; independent hosts are safer long-term.
Tasks
- Pick a host: Transistor ($19/mo, recommended), Buzzsprout ($12/mo, beginner-friendly), Captivate ($19/mo)
- Create the show in the host: title, premise, cover art, category, language
- Configure RSS metadata: explicit rating, copyright, category, type (episodic vs serial)
- Set up custom domain (podcast.yourdomain.com) for the show page
- Verify the RSS feed validates: paste into Cast Feed Validator
Pointers
- [Tool] Transistor.fm
- [Tool] Cast Feed Validator
[!CAUTION] Gotchas
- Anchor (Spotify) used to be free + popular; it locks your audio behind their platform and exits are messy. Pay $12-19/mo for an independent host.
- If you migrate hosts later, you have to set up RSS redirects. Transistor + Buzzsprout handle this cleanly; some hosts don't.
- iTunes-deprecated namespace tags (itunes:author, itunes:summary) are still required by Apple. Hosts handle this; verify in the validator.
Step 8: Submit to Apple Podcasts, Spotify, Overcast, and Pocket Casts
Estimated time: 1-2 hours active, 1-7 days waiting
Submission to directories is one-time per show; episodes auto-flow via RSS afterwards. Apple Podcasts approval takes 1-3 days; Spotify is instant; smaller directories take 1-7 days. Most modern hosts auto-submit on your behalf. The only platform that needs manual handling is YouTube Podcasts (now significant).
Tasks
- Submit to Apple Podcasts via Podcasters Connect: paste RSS, fill metadata, submit for review
- Submit to Spotify for Podcasters: paste RSS, claim show, instant
- Submit to Pocket Casts, Overcast, Castbox via your host's auto-submit (or manually)
- Submit to YouTube Music + YouTube Podcasts via YouTube Studio (manual, separate process)
- Verify the show appears in each directory's search 1-7 days post-submission
Pointers
- [Official] Apple Podcasts Connect
- [Official] Spotify for Podcasters
[!CAUTION] Gotchas
- Apple Podcasts review rejects shows with broken RSS, missing artwork, < 1 episode, or copyright violations in the music. Validate before submitting.
- Spotify's instant approval doesn't mean your show is searchable instantly — search indexing takes 24-72 hours.
- If your show name has trademark conflicts (similar name to existing show), Apple silently rejects. Pick a name that's findable + non-conflicting.
Step 9: Ship the launch and the first 3 episodes
Estimated time: 1 day live + 1 week follow-up
Launch day: trailer + episode 1 + episode 2 + episode 3 all live. Three episodes at launch dramatically outperforms one episode at launch — listeners who like episode 1 immediately have 2 more to consume, which is the strongest subscriber signal. Cross-promote on every channel: X, LinkedIn, email list, communities, your own newsletter.
Tasks
- Day-of: confirm trailer + episodes 1-3 are live in Apple Podcasts + Spotify
- Send the launch announcement: X thread, LinkedIn post, email blast, Discord / Slack communities
- Post the show + first 3 episodes URLs to relevant subreddits (no spam — community-friendly)
- Ask 5-10 close friends to subscribe + leave a review (reviews = ranking signal)
- Set the next 3 weeks' ship cadence: episode 4 in week 2, episode 5 in week 3, etc.
[!CAUTION] Gotchas
- Launching with 1 episode + a trailer caps listener engagement. 3 episodes at launch is the lift.
- Don't ask for reviews — Apple's algorithm penalises detection of organised review-asking. Ask friends to subscribe; reviews come naturally.
- Cross-posting to subreddits as 'check out my new podcast' gets removed. Engage in the community for weeks first, then mention the podcast in context.
Step 10: Build the agent-assisted episode pipeline
Estimated time: 1-2 days setup, ongoing
Each episode generates 5-10 pieces of distribution content (show notes, transcript, episode description, X thread, LinkedIn post, audiograms, blog post). Without automation, a weekly podcast becomes a 12-hour-per-week project; with agents drafting, it drops to 4-6 hours. Wire the agent to draft show notes from the transcript, episode description from the show notes, social copy from the description.
Tasks
- Get auto-transcription from your host or Descript (most provide this)
- Configure the agent to ingest the transcript and draft: timestamped show notes, episode description, X thread, LinkedIn post
- Define the agent's voice rules: matches your show voice, no hyperbole, lead with the hook
- Test on 2-3 past episodes; iterate the prompt until drafts are approval-grade
- Per-episode flow: record → upload to host → agent ingests transcript → drafts in Episode pipeline row → human reviews + ships
[!CAUTION] Gotchas
- AI-drafted show notes that read as 'In this episode, we discuss X and Y and Z' read robotic. Train the agent to lead with the hook.
- Auto-transcription has 5-15% error rate on names + technical terms. Spot-check before publishing.
- Don't publish the full transcript on day 1 — leak it to subscribers via email a week later as a reason to stay subscribed.
Agent prompt for this step
Process a new podcast episode and draft the distribution content.
Inputs:
- Episode transcript (from Descript / Riverside / host auto-transcription)
- Episode title + guest from Episode pipeline row
- Show premise + voice rules from Brief
Output:
1. **Show notes** (200-400 words):
- Bullet list with timestamps every 3-5 min
- Each bullet: timestamp + 1-line summary of that segment + key quote if any
- 3-5 internal links to related episodes
2. **Episode description** (200-400 chars): hook + premise + ends with "listen + subscribe"
3. **X thread** (5-7 tweets): hook tweet quoting the most-shareable line, body tweets with key insights, CTA tweet with episode link
4. **LinkedIn post** (1200-2000 chars): founder-story shape, references the episode and the most-shareable insight
5. **3 audiogram clips**: timestamps for the 3 most-shareable 30-60 sec segments (for video audiograms)
6. **Blog post outline** (8-12 H2/H3): for the optional canonical written companion piece
Constraints:
- Match the show voice (irreverent / serious / technical based on Brief)
- No 'we are excited to share', no marketing fluff
- Lead with the most surprising insight, not the most expected
- Internal-link 3-5 related episodes from Episode pipeline
Output as a structured update to the Episode pipeline row.
Step 11: Monitor downloads, iterate on what works
Estimated time: Ongoing, 1-2 hours/month
Podcast analytics are sparse: downloads (the only meaningful metric), completion rate (if your host provides it), and Apple Podcasts subscribers. Track which episodes outperform your baseline by 50%+ — those are the format / topic signals to repeat. Drop or evolve formats that consistently underperform after 5+ episodes.
Tasks
- Monthly: pull download numbers per episode from your host
- Track the rolling 30-day download average
- Identify the top 3 episodes (by downloads + completion rate) — what pattern? guest type? format? title style?
- Identify the bottom 3 — what to drop or evolve
- Update the next 5 episodes' planning based on the patterns
[!CAUTION] Gotchas
- Downloads compound on Apple's algorithm: Top Charts in your category drives 2-5x organic discovery. Episodes that hit Top Charts earn the show subscribers, not the episode.
- Completion rate is the algorithm's true signal. A 30 min episode with 80% completion outranks a 60 min episode with 40% completion.
- Don't quit before episode 25. Most podcasts hit a download plateau at episode 8-15 before compounding starts at episode 20+. Patience compounds; quitting compounds nothing.
Agent prompt for this step
Run the monthly podcast analytics audit.
Inputs:
- Per-episode downloads from the host (paste the numbers)
- Episode pipeline (titles, formats, guests, dates)
- Brief (premise, voice, format intent)
Output to the Brief as a markdown section:
1. **By the numbers**: 30-day downloads, total episodes shipped, average downloads/episode, top quartile vs bottom quartile
2. **Top 3 performers** + the pattern they share (guest profile, format, title style, topic)
3. **Bottom 3 underperformers** + the suspected cause (topic miss, guest fit, length, ship time)
4. **Format insights**: solo vs interview vs panel performance (if mixed)
5. **Title patterns**: which title structures earn the most clicks?
6. **Three concrete recommendations** for next month's episodes
Constraints: data-led. Don't celebrate vanity metrics; analyse what drove the downloads.
Hand the template to your agent
Paste the prompt below into your agent's permanent system prompt so the agent reads, writes, and maintains this workspace as you work through the steps.
You are an agent on the "Ship a podcast" playbook workspace at your-org/ship-a-podcast-from-idea-to-rss-feed.
Your role: own the episode pipeline + draft show notes + episode descriptions + per-episode social copy.
Cadence:
- For each new episode added: research the guest / topic, draft the question outline, draft a pre-show research brief.
- After recording: ingest the transcript, draft show notes (3 bullets per timestamp), draft the episode description, draft 3 X/LinkedIn posts.
- Maintain the Episode pipeline: ship dates, edit status, directory submission status, download counts.
- Monthly: surface top-performing episodes + patterns to the Brief.
First MCP tool calls:
1. list_surfaces(workspace_slug="ship-a-podcast-from-idea-to-rss-feed")
2. list_rows(workspace_slug="ship-a-podcast-from-idea-to-rss-feed", surface_slug="episode-pipeline")
3. get_doc(workspace_slug="ship-a-podcast-from-idea-to-rss-feed", surface_slug="brief")
Constraints:
- Episode descriptions: 200-400 chars, lead with the hook
- Show notes: timestamped bullets every 3-5 min of episode
- Episode titles: 50-80 chars, lead with the topic, end with the guest name
- Always include 3-5 internal links to related episodes
FAQ
How long does it take to launch a podcast?
Setup takes 3-4 weeks: 1 week for premise + format + gear, 2 weeks for recording the first 3 episodes + the trailer, 1 week for cover art + host setup + directory submission. Apple Podcasts review adds 1-3 days; Spotify is instant. Then you ship weekly or bi-weekly; the production loop takes 4-8 hours per episode once dialled in.
Do I need expensive recording gear?
No. The starter kit that sounds great: Shure MV7 ($249) or Samson Q2U ($69) USB mic, Sony MDR-7506 ($99) closed-back headphones, $20 of acoustic foam, and a quiet room. That's $80-370 total. Spend the saved money on cover art design and editing software (Descript at $15/mo). A $2000 mic in an untreated room sounds worse than a $69 mic in a treated room.
What podcast host should I use?
Transistor ($19/mo, unlimited episodes) for most cases. Buzzsprout ($12/mo, has a free tier with 90-day retention) if you want to start free. Avoid Anchor (now Spotify for Podcasters Lite) because the audio gets locked into Spotify's ecosystem and migrating later is messy. Pay the $12-19/mo for an independent host that owns its RSS infrastructure.
Do I need to be on YouTube?
Increasingly yes. As of 2026, ~30% of podcast listening happens on YouTube (especially video podcasts). The pattern that works: audio-first show, simple video (talking head + show notes overlay), upload to YouTube Music + YouTube Podcasts via YouTube Studio. You don't need full video production; static frames or simple talking heads work.
How do I grow the audience?
Three multipliers: (1) be on other podcasts (guest appearances drive 3-10x more subscribers than your own marketing), (2) consistent ship cadence builds compounding discovery (Apple's Top Charts reward consistency), (3) one strong epoch-defining episode every 5-10 episodes drives word-of-mouth (the episode someone shares with a friend). Buying ads for podcasts has terrible ROI for new shows. Invest in the network instead.
Can my AI agents help with the podcast?
Yes, especially for: drafting show notes from the transcript, drafting episode descriptions, drafting per-episode social copy, identifying audiogram clips from the transcript, drafting the blog companion piece, running the monthly analytics audit. Agents are bad at: the actual conversation, the editorial judgement on which guest fits, the voice that subscribers recognise. Use agents for the surrounding production work; the host stays human.