A host asked me last month whether podcast SEO was real or just something agencies invented to sell retainers. Fair question. I had been staring at his back catalogue for a week, so I had an answer ready: it is real, it just does not work the way blog SEO works, and almost nobody is doing the part that actually moves.
Here is the short version of my position. Nobody types a keyword into Apple Podcasts and finds your show. But people do type things into YouTube, into Google, and now into AI assistants, and those three surfaces read your episode as text. If the text is thin, you are invisible. If the text is good, your episode gets found by people who have never heard your name. That gap is the whole opportunity, and it is mostly a production problem, not a marketing one.
Search does not find your podcast, it finds your text
An audio file is a black box to a crawler. It has a title, a description, and a date. That is the entire surface area you are giving search unless you deliberately create more.
Video changes that. YouTube auto-transcribes everything you upload, and that transcript is indexed. Google pulls video into results for questions. The moment your episode exists as video with real spoken content, you have gone from three fields of metadata to forty thousand words of indexed text. This is one of the practical reasons I keep pushing hosts toward video first, and why I wrote about choosing between YouTube and Spotify as your primary home rather than treating them as equals.
So the first move is not keyword research. It is making sure your show produces indexable artifacts at all.
The four artifacts that actually get indexed
For every episode we cut, there are four things that can rank. Most shows produce one and a half of them.
- The video itself, with its transcript, title, and description. This is the heavyweight.
- Show notes with timestamps and chapters. Chapters become clickable segments on YouTube and give Google discrete topics to pull from.
- The episode page on your own site, with the full transcript pasted in and cleaned up.
- Clips, each with its own title and description, each a separate entry in the index.
We sell show notes and chapters as a $35 add-on, and I will be honest about why the price is what it is: it takes a human who watched the episode maybe forty minutes to do properly. You cannot generate good chapters from a transcript alone because the transcript does not know which moment was the turn. Our editors watch the whole thing anyway, so they already know.
Your episode page with the full transcript on it is the cheapest SEO asset in podcasting, and roughly nine out of ten shows I look at do not have one.
Titles are the highest leverage thing you edit
I have watched hosts spend three hours on a cold open and eleven seconds on the title. The title is the only part of the episode that most people will ever read.
The thing that broke my brain when I started paying attention: the best performing titles are almost never the topic. They are the claim. "Marketing for Founders" is a topic. "We spent $40,000 on ads and got nine customers" is a claim. Search rewards the second one because the second one matches how people actually type when they have a problem.
A working rule I use in edit notes now:
- Lead with the specific noun, not the category. "Cold email" beats "outbound".
- Put a number in if a real number exists in the episode. Do not invent one.
- Guest name goes at the end unless the guest is the search term. Most guests are not the search term. Some are, and then it goes first.
- Under 60 characters if you can, because the tail gets cut on mobile.
Then the thumbnail carries the click that the title earned. Those two are one unit, and I stopped pretending otherwise. Our sister studio ClickTheory does the thumbnail side because I got tired of watching good titles die under bad art.
Clips are search too, and most people forget that
Everyone treats clips as a social play. Reach, algorithm, discovery, all of that. True. But a clip on YouTube is also a permanent indexed page with its own title, and it competes for long tail queries that your full episode is too broad to win.
Ninety minutes of conversation cannot rank for a narrow question. A ninety second answer to that exact narrow question can. This is why I think clips belong at the top of the funnel rather than at the end of the production line, and why the pack of vertical clips is the add-on hosts renew most often.
The mistake I see is naming clips as fragments. "Ep 42 clip 3" is not a title. "The one question that kills most discovery calls" is a title, and it will still be picking up views in eighteen months.
Consistency beats optimization, and it is not close
Here is the blunt part. If your show publishes when you feel like it, no amount of keyword work will save you. Search surfaces reward supply. A show that posts every Tuesday for a year builds a body of indexed material that a show posting five times in a year cannot touch, even if the second show has better titles on every single one.
This is boring advice, which is why nobody sells it. It is also why our packs exist. The 4-episode pack saves 15% and the monthly saves 25%, and the discount is not generosity. It is that a host committing to a cadence is a host whose show gets easier to edit, and whose back catalogue starts compounding.
First cut lands in 48 hours, revisions come back the next working day, and if a launch date is tight there is 24-hour rush at $75. That matters for SEO more than it sounds like it should, because episodes that slip a week tend to slip two, and then the cadence is gone.
Where podcast SEO stops working
I am not going to pretend this is a growth channel for everyone. It is not.
Podcast search works when people are already looking for what you talk about. If your category has real search demand, questions people type, problems they name, then a well produced back catalogue is a slow, compounding asset. If your category has no search demand, you are optimizing for an audience that does not exist yet, and you would be better off spending that energy on direct outreach or paid. I said most of this already in the piece on when a podcast is the wrong channel entirely, and I still believe it.
The other failure mode is splitting your show before it has weight. Two shows means half the cadence on each, half the indexed material, half the authority. If you are thinking about that, read what I think about launching a second channel first, because the SEO math is usually against it.
The production choices that quietly help ranking
Some of this lives in the edit, which is the part nobody frames as SEO.
Audio quality is a ranking factor by proxy. People do not watch what is unpleasant to hear, watch time drops, and the algorithm reads that as a bad video. This is the honest case for a human on the audio pass, and I laid out the tradeoff in detail when comparing automated processing against a human mix. Automation gets loudness right. It does not know that the guest's best line was said quietly.
Pacing is the same story. Retention is the input search cares about most, and retention is decided by whether the first two minutes justify the next fifty. Our editors cut with that in mind rather than trimming silences on a threshold, because a threshold does not know what a pause means.
And structure helps directly. An episode that moves through distinct segments produces clean chapters. Clean chapters produce timestamps. Timestamps produce the little jump links Google sometimes shows under a video result. It is a small thing that only exists if the edit had a shape in the first place.
What I would actually do first
If you have thirty episodes sitting there and you want to start, do it in this order.
- Rewrite the titles on your ten best episodes. Free, takes an afternoon, and it is the only lever with immediate effect.
- Get transcripts onto your own site as episode pages. One page per episode, full text, real headings.
- Add chapters going forward. Do not retrofit the whole catalogue, it is not worth the hours.
- Cut three clips per new episode with standalone titles.
- Then, and only then, worry about keywords.
Most hosts do that list backwards, starting with a keyword tool and never getting to the transcripts.
Podcast SEO is not a myth. It is just slower and more boring than the people selling it want you to believe, and it lives in production decisions rather than in a plugin. If you want a hand with the editing side of it, whether that is how we run an episode end to end or just a sanity check on your setup, write to me at samy@pixelsamy.com. I read all of it myself.