Spotify or YouTube: Which Platform Actually Gives a Podcast Guest Search Visibility?

What we measured, and why text is the unit

We fetched show pages on each major platform with a desktop browser user agent, stripped script and style blocks, and counted the words a crawler would actually receive. Directory URLs were discovered through Apple's public search endpoint rather than constructed, because a guessed URL measures the guess.

Text is the right unit because it is what search and AI systems read. As documented elsewhere in this series, audio files are opaque to anything that does not transcribe them, and Google's own condition for appearing in its AI features is that a page be indexed and eligible to show a snippet.

The results separate the platforms sharply. Apple Podcasts returned 6,045 words for Darknet Diaries, 6,872 for Acquired, 7,258 for Freakonomics Radio, 5,978 for Talk Python and 4,339 for Accidental Tech Podcast.

Spotify returned four words. Not four hundred — four. We tested four separate show pages and got the same result each time, which makes it a property of how Spotify renders pages rather than an accident of one URL.

It is worth being precise about what a word count is measuring. We are not judging page quality or how useful the page is to a person. We are measuring how much text arrives in the HTML response, because that is the raw material available to any system that reads pages without running a browser — which includes a substantial share of crawlers and retrieval pipelines.

The spread within a single platform is narrow, which is what makes the comparison meaningful. Apple returned between 4,339 and 7,258 words across five unrelated shows, and Spotify returned exactly four on every show tested. Consistency at both ends suggests we are measuring a platform property rather than the peculiarities of any one page.

What the Spotify result does and does not mean

It does not mean Spotify is useless, and it does not mean nobody can read the page. A browser executing JavaScript renders the show page normally, and a crawler capable of rendering may see the content too.

What it means is that anything reading raw HTML — which includes a great many crawlers, scrapers and retrieval pipelines — receives almost nothing. The page exists and says four words.

There is a second layer. Spotify's robots.txt blocks GPTBot outright, as does Amazon Music's. So for at least one major AI crawler, the question of rendering does not arise: it is not permitted to fetch the page at all.

For a guest, this is the practical consequence. A Spotify listing is a real destination for listeners and a poor one for machine discoverability, and those two facts are routinely conflated when platform reach is being sold.

The asymmetry also explains a common experience. A guest checks their Spotify listing, sees a complete page with episode descriptions and artwork, and reasonably concludes the content is public and indexable. Both impressions are correct from a browser and neither describes what a plain crawler receives.

Why YouTube behaves differently

Because on YouTube a podcast is not an audio file with a page attached — it is a video with a transcript.

YouTube's documentation on RSS delivery states that it "will create videos for each podcast episode that you choose to upload", using the show art to build a static-image video, and that new feed episodes upload automatically.

The consequence is that an audio-only show acquires captions. YouTube generates automatic captions in more than sixty languages, warns that they "might misrepresent the spoken content" and tells creators to "always review automatic captions and edit any parts that haven't been properly transcribed".

That produces a searchable text layer for an episode whose own website may carry none. In our episode-page measurement, several shows published fewer than a thousand words per episode while their YouTube presence carried full machine-generated captions of the same conversation.

The caption layer is machine-generated, which cuts both ways. It exists without the host doing anything, which is why it covers shows that publish no transcript of their own — and YouTube itself warns the output may misrepresent what was said. For a guest whose company name or technical term is mangled by speech recognition, the searchable text is present and wrong.

Do the directories themselves permit crawling?

Mostly yes, with two exceptions that matter.

We fetched robots.txt for eleven podcast destinations. Apple Podcasts, Pocket Casts, Podchaser, Listen Notes, Castbox and YouTube all carry narrow path rules — administrative directories, search endpoints, user areas — rather than blanket restrictions on content. iHeart's wildcard group carries no disallow rules at all.

Spotify and Amazon Music each block GPTBot by name. Overcast returned no readable robots.txt and Player FM refused our requests, so both are unknown rather than open.

This matters for the common claim that a show appears on twenty platforms. A destination that renders client-side, or blocks the crawler you care about, contributes a URL rather than discoverable text — and the two are not the same asset.

The two exceptions are not marginal platforms. Spotify and Amazon Music are among the largest audio destinations in the market, and both decline the crawler most often named in discussions of AI visibility. Anyone selling AI discoverability through broad podcast distribution is selling reach across destinations, two of which have said no in writing.

So which platform is most valuable to a guest?

It depends on what the appearance is for, and the honest answer has two halves.

For reaching listeners, platform share is what matters, and Spotify and Apple dominate consumption regardless of what their pages return to a crawler. A guest who wants to be heard should not care about server-rendered word counts.

For being found later — by a search engine, an AI assistant, or a person who half-remembers your name — the asset is text on a crawlable page. On that measure Apple's pages carry the most, YouTube adds a caption layer the audio never had, and Spotify contributes the least despite being the most prominent destination.

None of which is a criticism of Spotify's product. It is a client-side application that happens to have URLs, and it was not built to be a text archive.

The two halves also decay differently. Listening happens in the first days after publication, as the traffic report elsewhere in this series documents, while a crawlable page carrying the conversation keeps working for as long as it remains indexed. The platform that matters for the launch and the platform that matters a year later are rarely the same one.

Both answers are worth having before you choose where to push an appearance.

What this means for anyone booking appearances

Stop treating platform count as reach. Appearing on eight directories produces eight listings of varying readability, not eight sources of discovery.

Prioritise the destinations that publish text. Apple's pages carry substantial content, YouTube generates captions for audio-only shows, and the show's own episode page is the one place a guest can sometimes influence — by asking whether a transcript will be published.

Check the crawler rules for any platform you are counting on. They are public, they take a minute to read, and two of the largest audio destinations block a major AI crawler outright.

Media Morpheus books podcast appearances. What a booking delivers is an appearance before a real audience, plus whatever text the ecosystem happens to publish around it. The second part varies by two orders of magnitude between platforms, and nobody mentions that when selling distribution.

Methodology and limitations

This is an original measurement conducted in September 2026. Show pages were retrieved with a desktop browser user agent; script and style blocks were removed before counting words. Directory URLs were discovered through Apple's public search endpoint rather than constructed. robots.txt files were retrieved separately and parsed by grouping consecutive user-agent lines so each named agent was assessed against the rules actually governing it. Six limits apply. First, word counts measure server-rendered HTML only: Spotify and other client-side applications render content in a browser that a plain crawler does not receive, so the figure describes what raw HTML contains, not what a human or a rendering crawler sees. Second, this is a purposive selection of established shows, not a representative sample. Third, Overcast returned no readable robots.txt and Player FM refused our requests; both are recorded as unknown. Fourth, an earlier measurement in this work incorrectly recorded an Overcast page that was in fact a login page; that figure was discarded and is not reported. Fifth, robots.txt records stated permission rather than crawler behaviour. Sixth, all figures describe a single moment and these pages change.

Conclusion

Asking which platform gives a guest search visibility turns out to have a measurable answer, and it is not the one the industry's platform-count marketing implies. Apple's show pages returned around six thousand words of readable text; YouTube's returned roughly two thousand and add machine-generated captions to audio that had none; Pocket Casts returned a few hundred; and Spotify returned four, across every show we tested, while separately blocking GPTBot. For listening, Spotify is indispensable. For being found afterwards by anything that reads text, it contributes almost nothing — and a guest choosing where to push an appearance should know that those are two different questions with two different answers.

Frequently Asked Questions

Sources

1. vendor research Media Morpheus, September 2026. Original measurement of podcast platform pages and robots.txt files. Method and limitations stated above.
2. official Google Search Central. AI features in Google Search, for the indexing and snippet eligibility condition.
3. official YouTube Help. Deliver podcasts using an RSS feed, for how YouTube converts audio episodes into videos.