How to write podcast show notes from a video recording

An episode does well. Downloads are up, the guest shared it, the numbers in the dashboard look like progress.

The episode page on the show's own site ranks for nothing. It contains an embedded player, a title, and two sentences written in a hurry on publication day. To a search engine, that page is almost empty, because the ninety minutes of substance inside it are audio, and audio is not text.

This is the single largest gap between how much value podcast produces and how much of it is discoverable.

What should podcast show notes include?

Usable show notes contain a short summary of what the episode covers, chapter markers with timestamps, the guest's full name and how they should be described, every book, product, study, or link referenced during the conversation, and, where possible, the full transcript. Industry guidance converges on roughly this list; Podcast.co's breakdown of what belongs in show notes is a representative version. The summary serves the listener deciding whether to press play. The transcript and the specific references serve everyone who was never going to find the episode without them.

Audio is invisible; the page is what gets indexed

The mechanics are worth stating plainly because a lot of podcast marketing advice skips them.

A search engine indexes the episode page. It reads the title, the description, and whatever else is written there. It does not listen. Whatever brilliant explanation of a topic occupies minutes forty through fifty-two of the episode contributes nothing to that page's ability to answer a query about that topic.

Two sentences of description gives a search engine two sentences of signal. A transcript gives it the entire vocabulary of a ninety minute expert conversation, including the specific phrases that people actually type: the name of a technique, an error message, a book title, a niche term a guest used four times.

That is the whole mechanism. Everything else about podcast discoverability is downstream of it.

Video podcasts changed the problem in both directions

Most shows now record video, and the consequence for discoverability cuts two ways.

The upside is a second distribution surface with its own search behavior, and a video file that already contains everything needed to produce written material.

The downside is that the video platform captures the text work. Chapters get written into the video description, the auto-generated captions live on the platform, and the show's own episode page stays as thin as it ever was. The show has done the work and given the benefit to somebody else's domain.

The correction is not to stop publishing video. It is to treat the video file as the source for the show's own written page, which is the asset the show actually owns.

What belongs on the episode page

Six components, in descending order of how much they are worth relative to the effort.

A transcript. The single highest-value element and the one most often skipped. It is also an accessibility measure, which matters independently.

Chapters with timestamps. These serve listeners who want one segment, and they give the page an internal structure that makes it readable rather than a wall.

The guest, named correctly and described accurately. Spelling matters here more than anywhere else on the page, because the guest's name is a search term and a misspelled name is invisible.

Everything referenced. Books, papers, tools, companies, other episodes. Guests mention these constantly and listeners ask for them constantly. Compiling them is ten minutes of work that answers most of the show's inbound email.

A summary that is not a teaser. Marketing copy describing how fascinating the conversation was performs poorly. A summary stating what was actually claimed performs well, because it matches what people search for.

One or two quotable passages, verbatim. These get lifted into social posts and into other people's writing, and they need to be accurate.

Why Markdown is the right intermediate format

The episode page will end up in a content management system, a static site, or a hosting platform's editor. Producing it in Markdown first avoids most of the friction.

Markdown is plain text with a small, specified set of conventions, pinned down by CommonMark into an actual specification with a reference implementation. Practically, that means headings survive as headings, timestamps stay legible as literal characters, and the file pastes into almost any editor without carrying invisible formatting with it.

It is also the format most easily reused. The same file supplies the site page, the RSS description, the newsletter, and the video platform's chapter list, without reformatting each time.

Producing them from the recording

The reason show notes are thin is not that podcasters disagree about their value. It is that writing them from scratch after publishing an episode takes an hour that does not exist.

The first draft is worth automating. Converting an uploaded video to markdown returns the episode already divided into timestamped chapters with the speakers separated, which for an interview show is the structural work almost complete: host and guest distinguished, topic boundaries marked. Vomo also generates a summary with the key points from the recording and exports to Markdown, plain text, Word, or PDF, so the file goes straight into the site's editor.

Speaker separation is the feature doing the most work here. An undifferentiated transcript of a two-person conversation is close to unusable for pulling quotes, because there is no reliable way to tell which sentences belong to the guest, and the guest's sentences are the ones anyone wants to quote.

What a person still has to fix

Automatic output is a draft. Four passes make it publishable, and they take about twenty minutes.

Names, always. Guest name, company names, product names, and any person mentioned in passing. These are the least reliable part of any automatic transcript and the highest cost when wrong.

Numbers and claims. Any statistic or date the guest cited should be checked before it appears in written form on the show's own domain, because a written page invites scrutiny that a spoken remark does not.

Chapter titles. Automatically generated chapters land in approximately the right places with approximately the wrong wording. Rewriting them into the language a listener would search for is where most of the search benefit comes from.

Ad reads and sponsor segments. These generally should not appear in the transcript on the page, and removing them is a deliberate decision rather than an oversight.

The guest has a reason to share a good page and none to share a bad one

There is a distribution argument that gets overlooked because it sounds like a courtesy rather than a tactic.

Guests promote episodes unevenly, and the variable is not how much they enjoyed the conversation. It is whether there is something worth linking to. A page with the guest's name spelled correctly, their title and affiliation right, their book or company linked, and two accurate quotable passages is a page they will send to their own audience, because it reflects well on them.

An episode page with an embedded player and two sentences gives them nothing to send. The link goes out once, on the day of publication, and stops.

This compounds with booking. Prospective guests look at how previous guests were presented before agreeing to appear, and a show with substantial written pages reads as one that takes the conversation seriously. The show notes are, in that sense, part of the pitch.

Three specifics make the difference in practice. Confirm the guest's preferred name and title rather than lifting it from a bio written four years ago. Link the thing they are currently promoting, not the thing they are best known for. And send them the page before it goes live, which takes two minutes and reliably produces a correction worth having.

The archive effect

The argument for doing this consistently rather than occasionally is cumulative.

A show with two hundred episodes and thin descriptions has two hundred pages competing for the same handful of generic queries. The same show with full written pages has two hundred pages each capable of matching a different specific query, and the back catalog starts producing traffic years after publication.

It also becomes searchable internally, which matters more than expected. Producers routinely need to find the episode where a particular thing was discussed, and with text they can. Without it, the show's own team is in the same position as its audience: certain the conversation happened, unable to locate it.

The short version

A podcast produces a large quantity of specific, expert language and publishes almost none of it in a form anything can read.

The recording already contains everything the episode page needs. The gap between the two is one conversion step and about twenty minutes of editing per episode.

Downloads measure the audience a show already has. The written page is the only part that reaches the audience it does not.