Three assumptions run through most podcast SEO guidance and all three are unsafe. That there is a Google podcast index to get into, when that product shut down in 2024. That a technically correct page will be indexed, when my own clean site had four of its five articles skipped. And that the rich result you are marking up for will still exist, when Google has been steadily retiring them. What survives all three is the page and the transcript.
Search "podcast SEO" and some of what comes back will tell you to get your feed into Google Podcasts.
Google Podcasts does not exist. It was wound down across 2024: announced in September 2023, cut off for US listeners on 2 April 2024, discontinued worldwide that summer, with the creator-facing Podcasts Manager going dark on 29 July 2024. The app had passed 500 million downloads on the Play Store, and then it was gone. Both halves went, the listening app and the tool you would have used to submit a feed.
I ran that search while writing this. Near the top was a guide with 2026 in its title explaining how to submit your podcast to Google Podcasts. Not an old post that had drifted down the rankings. A current-dated guide, walking you through a process that has been impossible for two years.
Advice does not update itself, and dated titles are not evidence of dated checking. A page can carry this year's number and last decade's instructions.
That is the shape of the problem, and it goes deeper than one dead app. Most podcast SEO advice rests on three assumptions about how discovery works, and none of them is safe to build on.
Assumption one: there is a podcast index to get into
The mental model behind most guidance is that Google maintains a directory of podcasts, and your job is to get listed in it and rank inside it.
There is no such thing to get into any more. What exists is ordinary web search looking at ordinary web pages, plus a set of separate closed systems, Spotify and Apple and YouTube, each with its own internal search that does not read your website and does not care about your schema markup.
That split matters because it changes what you are optimising and for whom. Work you do on your episode page has no effect on where you appear inside Spotify. Work you do on your Spotify metadata has no effect on Google. Advice that blurs the two, and a lot of it does, will have you doing the right task against the wrong system.
Assumption two: technically correct means indexed
This is the one I got wrong myself, and it is the reason I trust the observation.
When this site went live it had been built carefully. One linked structured-data graph across every page, a clean sitemap with real modification dates, canonical tags, a valid robots file, fast pages, no crawl blocks. It passed its own audit and I had no reason to think anything was missing.
Google had indexed ten of the fourteen pages. Four of the five blog articles were simply not there.
Nothing was wrong with them. They were not blocked, not duplicated, not thin. Google had crawled the site and decided those particular URLs were not worth putting in the index yet, which is a decision it makes constantly and does not explain. They went in only after I submitted each missing URL by hand through Search Console.
None of the podcast SEO guides I read mentioned this step, because they describe how indexing is supposed to work rather than what actually happens to a new site with no authority behind it. If you have published thirty episode pages and assumed they are all in Google because your setup is correct, check. The number is often lower than you would guess, and the fix is unglamorous and manual.
Assumption three: the rich result you are marking up for will still be there
A large amount of SEO effort goes into structured data aimed at a particular visual treatment in the results page: the expandable questions, the step-by-step boxes, the enhanced listing.
Google has been retiring those, and the FAQ case is worth following all the way through because of how long it took and how quiet it was.
FAQ markup arrived in 2019 and was adopted widely across the web. In August 2023 Google narrowed it, restricting the rich result to well-known, authoritative government and health sites, which meant that from that point an independent show's site was not eligible regardless of how correct its markup was. Then on 7 May 2026 the FAQ rich result stopped appearing in Google Search entirely. The search appearance, the rich result report and the Rich Results Test support were dropped from Search Console in June 2026, and the Search Console API support went in August.
How-to went the same way: no longer shown on desktop, then deprecated, then the documentation removed. Reporting for a batch of other types, among them course info, estimated salary and special announcement, was pulled as well.
Seven years from launch to removal, with nearly three of those years in a state where most sites were marking up for a result they could not get.
None of this makes structured data pointless, and I still put a full entity graph on every page here. But the reason has changed. Markup is now worth doing because it makes a page legible to machines that need to understand what they are reading, which increasingly means answer engines rather than a blue-link results page. It is no longer a reliable route to a specific decoration in Google.
The practical version of that: if a piece of markup only earns its keep by producing a visual feature, treat the feature as temporary. On this site the questions and answers are rendered visibly on the page and generated from the same source as the schema, so they say the identical thing. When the rich result went away, the page did not lose anything, because a reader could always see the answers. That is the test worth applying to any markup you add. If Google removed the feature tomorrow, would the page still be better for having it?
What actually holds
Strip out everything that depends on a product staying alive or a feature staying supported, and a small durable set is left.
A page per episode, on a domain you control. Not a platform link, not an embed on somebody else's site. A URL that is yours and can be cited.
A full transcript on that page. It is the largest body of indexable text the episode will ever produce, it contains the phrasings people actually search for, and it is the part an answer engine can quote from. Everything else on the page is an accessory to it.
A title that matches how you would type the problem, not how the episode was named in the studio.
And a way for a machine to tell what the page is. Whatever form that takes in a given year, the underlying job stays the same.
Notice what these have in common. Every one of them is about the text around the audio, and every one of them is decided long before the transcript exists, because the transcript is a record of what the host said. Which is the argument I have made separately about doing the keyword work before recording rather than after. This route arrives at the same place from the other end: even if you ignore everything about pre-recording planning, the durable half of podcast SEO is still the text layer, and the biggest single piece of that text layer is decided at the microphone.
The indexing observation is one site, mine. It is a real, dated, checkable data point and I have described it exactly, but it is a sample of one and I have not run a study on how commonly new podcast sites have pages skipped. I have also not found a controlled comparison showing that any particular podcast SEO approach outperforms another. That literature does not appear to exist. The reasoning here is built on documented platform behaviour, Google's own published deprecations, and what happened on this site.
The out-of-date guide I described is one example I hit on one search on one day. I have not counted how much of the advice out there has the same problem, so read that as an illustration of the failure mode rather than a measurement of how common it is. The dead product behind it is not in question either way.
Most podcast SEO advice was written for a search product that has since been discontinued, and assumes a level of automatic indexing that a new site does not get.
The parts that survive are boring and they have not changed in years. Own the page. Publish the transcript. Title it the way a person would ask. Then go and check whether Google actually took it, because it may not have.