Spotify Podcast Episode Extractor

Export every episode of any Spotify podcast — title, full description, duration, release date, share link — and then get the things the description actually contains: chapter timestamps, sponsor domains, and the guest names in the title.

Built for media researchers, PR teams hunting interview slots, and content analysts tracking what podcasts are covering. A podcast's guest list, chapter breakdown and advertiser roster are all buried in one free-text blob; this turns that blob into columns.

No Spotify account, no API key, no browser.

---

What you get per episode

Field	Notes
episodeName, description, releaseDate, releasedAt	Spotify publishes release times to the minute
duration, durationMinutes	1:12:45 and 73
guestHints	Names read out of the title — a heuristic, honestly named
chapters, chapterCount	Every 12:34 Topic line, with second offsets
sponsorDomains	Advertiser domains, after stripping the show's own links
links, mentionedHandles	Everything else the description points at
hasVideo, hasTranscript, isPaywalled, explicit	What kind of episode it is
episodeUrl, shareUrl, coverArt, audioPreviewUrl	Links and media

Plus one SHOW row per podcast: publisher, description, episode count, average rating and review count, topics and cover art.

---

Quick start

{
  "showUrls": ["https://open.spotify.com/show/4rOoJ6Egrf8K2IrywzwOMk"],
  "showNames": ["Huberman Lab"],
  "releasedAfter": "2026-01-01",
  "maxEpisodesPerShow": 500,
  "exportFormats": ["csv"]
}

Podcasts can be given as links or plain names. Individual episodes go in episodeUrls.

---

The date window is also a cost control

Episodes arrive newest first, so releasedAfter does not just filter — it stops paging the moment the catalogue goes older than your date. Pulling September's episodes from a 2 753-episode show costs three requests, not twenty-eight.

---

About the mined fields

mineDescriptions costs no extra requests; it reads text already in the response. What it does is labelled honestly:

• chapters — timestamp lines in the description. A timestamp later than the episode's own duration is discarded, because it is a date or a price that looks like one.
• sponsorDomains — link domains with the show's own distribution and social platforms removed. In an ad-supported podcast these are the advertisers; in one that is not, they are just links.
• guestHints / episodeNumberHint — derived from the title. Nothing on Spotify marks who a guest is, so these are guesses from title convention. They are named …Hints for that reason and should not be treated as a field Spotify publishes.

---

Cost

One request for the show, plus one per 100 episodes. A 100-episode pull is three requests including the token. Looking a show up by name adds one; each individual episode URL costs one.

---

Limits worth knowing

• Transcript text is not public. hasTranscript tells you Spotify has one; the words are not exposed to an unauthenticated client and are not fetched.
• Paywalled episodes appear in the list with isPaywalled: true, but their audio is not accessible.
• episodeCount comes from the episode query, not the show query — Spotify's show metadata returns a single sample episode and no total.
• Descriptions vary enormously. Some publishers write full chapter lists; others leave the description empty. descriptionLength and chapterCount make that visible rather than silently returning nothing.

---

Output

Every run writes to the Apify dataset. Set exportFormats to also drop a ready-made spotify-podcast-episodes.csv, .xlsx, .json or .ndjson into the run's key-value store. Chapters and links are nested, so they are kept out of the spreadsheet columns and stay in the JSON exports.

See CRAWLING_METHOD.md for how the data is obtained.