He Did 2,000 Podcasts in a Year Just for SEO Rankings

Edward Sturm

Abstract

An hour on a podcast built as a search asset rather than as a show. Roughly two thousand episodes in a single year, each one aimed at a narrow buying query and syndicated across dozens of platforms, so that the same recording seeds the search results, the video carousel and the AI answers at once. The lever is not audience or entertainment — it is volume, relevance, and the repeated language that turns an episode into something citable.

• One search query is one episode, the way one query is one web page: five to fifteen minutes, single topic, no fluff.
• The bottom of the funnel is the target, which means an inquiry rather than a subscriber and a scoreboard that ignores watch-time.
• Syndication is a backlink harvest once the platforms are sorted into do-follow, no-follow and link-stripping.
• The claim goes in the first sentence as a clean triple and is repeated throughout, so the models cite it back.
• Retention is named the number-one video ranking factor, which is why uploading one episode to many channels costs reach.
• The biggest waste was the top of the funnel — there are far more buying-intent questions than expected.

Everything below follows the episode in order, from the two podcast models through syndication, entity and knowledge-panel work, into titles, descriptions and deliberate repetition, out through carousels, competitor targeting and press wires, and finally the economics: the team, the paid placements, and what the year cost. For anyone building a personal brand the useful part is the reframe — a recording is not content, it is a unit of search coverage.

Chapter summaries

00:00 - Podcast episodes can be produced at extreme volume purely as ranking assets

You can produce podcast episodes at extreme volume purely as SEO assets, not for an audience.

"You did 2,000 podcasts in a year for the purpose of SEO."

— 00:00

The number is the argument. Two thousand units in twelve months is not a show anybody could watch, which is the point: each episode is a ranking asset, and the catalogue is a coverage map of search demand rather than a body of work. Nothing about it is built to be enjoyed daily.

00:26 - Two opposite podcast models: a show people subscribe to, and episodes built to rank once

There are two opposite podcast models: an energetic show people subscribe to and watch daily, and keyword-specific episodes built to rank once and convert.

"It's been absolutely incredible. Um, it's been a lot better than what I thought it was going to be."

— 00:26

The split matters because it decides everything downstream. An audience model optimises for retention and return; a search model optimises for one query and one inquiry. Choosing which business you are in comes before any production decision, and the two rarely share a format.

01:58 - One search query is one episode, the way one query is one web page

Treat one search query as one episode the way you treat one query as one web page: focused, five to fifteen minutes, on a single topic.

"So my episodes are very much going after a search query that someone might want to search."

— 01:58

The discipline is the same one that governs a web page. One intent, one asset, five to fifteen minutes, no wandering. A sprawling two-hour conversation covering six topics is harder to rank for any of them than six short episodes each aimed at one.

03:18 - A keyword episode is bottom of the funnel — an inquiry rather than a subscriber

A keyword episode is bottom-of-funnel: the viewer watches once and turns into an inquiry, not a subscriber.

"But it's bottom of the funnel. I'm going to get an inquiry."

— 03:18

It reframes what success looks like. Nobody subscribes; somebody enquires. The viewer arrives at the moment of intent, gets the answer, and converts — which means watch-time, return rate and every other audience metric are the wrong scoreboard.

03:40 - "How to do X" is a top-of-funnel search, and video wins on "best X for Y"

"How to do X" is a top-of-funnel search; "best X provider for Y" is bottom-of-funnel — and video works especially well for the bottom.

"It was a top offunnel search not like a bottom ofunnel search."

— 03:40

The phrasing is the tell. A how-to query is somebody learning; a best-provider-for-a-situation query is somebody buying. Video ranks unusually well on the second kind, which is where the whole approach concentrates its effort.

04:15 - Video feeds the language models, and syndication manufactures consensus

Video is the leading content type feeding the large language models, and syndicating an episode like a press release builds consensus — among the biggest ranking factors for AI answer engines.

"anything you can think of at the bottom of funnel which would turn into an inquiry, we want to put a video format in there."

— 04:15

Two claims stack here. Video is the leading content type feeding the language models, and syndicating one episode across dozens of platforms reads as agreement rather than as one opinion — repetition across sources is the signal, not the volume on any single one.

06:04 - The syndication engine sorts platforms into do-follow, no-follow and link-stripping

The syndication engine: upload the video, push it through a podcast host to dozens of platforms, sort them into do-follow, no-follow and link-stripping, and force-index the useful ones.

"we get the video, we upload it onto YouTube, we then put it onto Transistor."

— 06:04

The mechanics are ordinary and the sorting is the work. Upload, push through a podcast host to dozens of platforms, then classify what comes back into do-follow, no-follow and link-stripping, and force-index the ones worth having. Syndication becomes a backlink harvest only after that classification.

06:48 - A media-database profile earns backlinks and strengthens the entity

Republishing the episode on a film-and-media database, and creating guest profiles there, earns backlinks and strengthens the entity.

"we would then get that video and publish it on IMDb where I had the guest on IMDb."

— 06:48

An authoritative third-party profile does two jobs at once: it carries a link, and it repeats who you are in somebody else's voice. The second job is the one that feeds the knowledge graph, and it is the one most link-building never touches.

08:30 - The "official websites" field places do-follow links and repeats the identity

A media-database profile's "official websites" field lets you place do-follow links to any site you want to rank, and repeats who you are — which strengthens knowledge-graph IDs and knowledge panels.

"Well, you can get back links. So if anyone wants to upload their podcast when you start uploading it in the edit edit part in the back end there's a place called official websites."

— 08:30

The field is the mechanism. A profile's official-websites slot places do-follow links at whichever pages need them, and the surrounding profile restates the identity. Repetition from an authoritative source is what raises an entity's confidence score.

09:39 - The traffic-tier method applies to video: win the reachable terms first

Apply the traffic-tier / cornerstone method to video: go after keywords you can rank for now, earn traffic, then internally link up to harder terms.

"I think people try to over complicate things and I think if you do cornerstoneing um you rank for you go after keywords that is within your bucket."

— 10:14

The method is not new, only newly applied. Go after the terms you can win now, earn the traffic, then internally link upward to the harder ones. Applying it to video changes nothing about the logic — the sequencing is the strategy.

10:56 - A do-follow description link on a high-authority platform ranks fast

A do-follow link in a description on a very-high-authority podcast platform, pointing to a hyper-relevant page, ranks fast because the crawler visits those sites constantly.

"all DR80 websites you it's difficult to get links on these websites and you're able to put a do follow back link in that description."

— 11:10

Three things compound: very high domain authority, exact relevance, and a crawler that visits constantly. A description link on that kind of platform is crawled almost immediately, which compresses the weeks a normal link takes into days.

11:52 - Low search volume is fine at the bottom of the funnel

Low search volume is fine at the bottom of the funnel — thirty to forty searches a month for a specific service and location still converts — so replicate the "service for industry in location" pattern across every industry and location.

"they might only get like 30 or 40 searches a month, but it's 30 40 searches bottom of the funnel."

— 11:52

Thirty or forty searches a month sounds like nothing until you multiply it by every trade and every town. Each query is tiny and high-intent, and the pattern replicates across a matrix no single broad term could cover.

12:29 - The episode is structured around the questions a buyer would actually ask

Structure the episode around the entity attributes a buyer would ask — the people-also-ask questions — then close with a soft pitch and a free-quote call to action.

"we'd think about the entity attributes like what questions would we want to."

— 12:27

The structure comes from the buyer, not from the format. Ask the questions a buyer would ask — the people-also-ask set — and close with a soft offer. The questions are the outline, which is why the episode needs no script.

14:03 - The backstory comes out and one line of credibility stays in

Cut the fluff and the backstory; jump straight into the topic and give only enough credibility to earn the listener's trust.

"I don't want any fluff. I don't want us going off subject."

— 14:03

Off-topic introduction is dilution. One line of credential establishes trust, and everything after it belongs to the topic. The backstory serves the guest; the searcher wants the answer, and relevance is measured against the searcher.

15:15 - Titles and descriptions are built the way a web page is built

Build titles and descriptions the way you build a web page: mirror people-also-ask, apply the "would this query open a new page?" test, and use the H2 questions as the questions you ask on camera.

"you would do it on your own website. So, um, if you're looking, let's say, people also ask, right?"

— 15:23

The transfer is one-to-one. Mirror the people-also-ask phrasing, apply the test of whether a query would justify a new page, and use the resulting questions as the on-camera questions. Proven on-page logic does not stop working because the asset is a video.

17:15 - The keyword belongs in the title field and in the first line of the description

Put the target keyword in the title field and in the first sentence of the description, then use an AI assistant on the transcript to write a bulleted description under the platform's character cap.

"we have the keyword in the title field and then at the beginning of the description field."

— 17:15

Front-loading is the whole trick. Keyword in the title field, keyword in the first line of the description, then a long bulleted description generated from the transcript and checked, because generated descriptions hallucinate. The check is not optional.

18:06 - The claim goes in the first sentence as a semantic triple the models can cite

State the claim you want cited in the very first sentence and as the title, then repeat it throughout as a semantic triple — subject, "says", claim — close to keyword stuffing, so the models pick it up and cite it.

"so let's do a real life example, right?"

— 18:06

The form matters as much as the frequency. Subject, verb, claim — a clean triple, stated first and repeated — is exactly the shape a model can lift and attribute. Saying it the same way every time is what turns repetition into citation.

20:05 - The repetition is deliberate, so frequency reads as agreement

The repetition is deliberate: use the target keyword and claim over and over in the transcript so the model treats it as cited by multiple sources.

"Are you using the keyword over and over again in the transcript deliberately?"

— 20:05

It is stated plainly as practice rather than dressed up as technique. The keyword appears religiously throughout the transcript, which is near enough to keyword stuffing that it gets named as such. Frequency of a well-formed claim is read as multiple sources agreeing.

20:37 - Generated video is not yet faster, and a black screen with text still ranks

An AI avatar and cloned voice can do this, but at present it takes the team longer than filming, and the phonetics sound robotic — while a black screen with text and the right transcript still ranks and feeds the models.

"could somebody take your your strategy and just do it with an AI?"

— 20:50

The constraint is practical, not principled. A cloned avatar and voice currently take the team longer than filming does, and the phonetics give it away on acronyms. Meanwhile the ranking value sits in the transcript and the syndication, so a black screen with text still works.

22:29 - The videos are built for the searcher rather than for the algorithm

The videos are built for a bottom-of-funnel searcher who wants to grow a business, not for the algorithm or for subscribers.

"yes, it's not entertaining, right?"

— 22:29

Everything follows from who the asset is for. The videos are not entertaining, not built for recommendation, and not aimed at subscribers. The person they are built for is searching a buying term and wants the answer, and optimising for anything else costs relevance.

23:20 - Naming rivals strengthens relevance rather than weakening the pitch

Keep the pitch balanced: give the searcher all the information, tell them to track KPIs and ROI, mention alternatives like other channels, and even name competitors — because those are semantically relevant entities for the query.

"However, make certain you track your KPIs. Make sure you're looking for a return on investment."

— 23:20

It reads as honesty and works as relevance. Telling the searcher to track return, naming other channels, even naming rivals — these are semantically related entities for the query. The even-handedness that makes the pitch credible is the same thing that makes the page relevant.

24:08 - The same episode uploaded to several channels fills more of the results page

Upload the same episode to multiple video channels — not marked as collaborators, with different titles and thumbnails — so the platform does not treat it as one piece.

"they really are not made to get pushed by like YouTube's algorithm."

— 24:08

One asset presented as several fills more of the page. Different titles, different thumbnails, no collaborator marking, so the platform does not collapse them into one. Occupying more of the result is the aim rather than reaching more people.

24:53 - The cost of that duplication is reach, because the signals split

The cost of that duplication is reach: user signals split across channels, so each channel's click-through and algorithmic push suffer.

"that's going to hurt reach because your user signals are going to be split between all these different podcast"

— 24:51

There is a bill for it. User signals split across channels, so no single upload accumulates the click-through and retention a recommendation system rewards. That is acceptable if rankings are the goal and unacceptable if reach is.

26:18 - Audience retention is the number-one video ranking factor

High audience-retention rate is the number-one video ranking factor, for both recommendation and ranking.

"So high high retention rate of views is the is the number one ranking factor um bar none."

— 26:18

Of everything named in the episode, this is the single-factor claim. Retention is rated the top ranking factor without qualification, for both recommendation and search, because it is the clearest proxy for whether the video answered the query.

27:34 - A press release is duplicate content by definition, which is the defence

A press release is by definition duplicate content — the same article syndicated across many channels — so treat cross-channel republishing as press-release behaviour rather than spam.

"some people go, it's duplicate content. Well, you tell me, what a press release is."

— 27:34

The defence is a definition. A press release is the same article syndicated across many outlets, and nobody calls that spam. Cross-channel republishing is the same behaviour under a different name, which moves the argument from ethics to precedent.

28:17 - Near-identical titles across channels trigger a video carousel

Uploading near-identical-title videos across channels triggers a video carousel in search: one lone video may sit around position eight, but several can capture three of the four carousel slots.

"when you do a very similar title video and you do it on multiple channels that triggers a video carousel."

— 28:17

The redundancy converts into placement. One lone video may sit around position eight; several near-identical ones can take three of the four carousel slots. It varies by keyword, but the mechanism is that a carousel rewards having more than one match.

29:26 - Competitor brand names are a video target, because video sits under their own site

Target competitor brand names with video — "competitor alternatives", "competitor vs [you]" — because video ranks in a carousel just below the competitor's own site.

"if you search maybe like a competitor brand name if there is video about that competitor, that video will show right below their website."

— 29:26

Comparison searches are buying searches. Somebody typing a rival's name with 'alternatives' has already decided to switch, and video sits in a carousel just under the rival's own site. Meeting the buyer at the comparison is cheaper than outranking the brand itself.

30:56 - The honest comparison still captures the carousel and still converts

Do the competitor comparison honestly: tell searchers to try the rival too and track ROI on both — it still captures the carousel and generates inquiries.

"if you go into the videos tab, you'll see maybe five or six of the videos being us."

— 30:56

Honesty is the version that survives. Tell searchers to try both and track the return, speak well of the competitor, and the carousel slot and the inquiries still arrive. The tactic does not require the bad-faith comparison it is usually paired with.

32:18 - Only one engine reads the transcript, so the script gets published as text

Only the search engine's own AI reads video transcripts; other assistants see just the description — so publishing the full script on the podcast page and syndicating it gets the words in front of the other engines.

"I found that Gemini is the only large language model that sees YouTube transcripts."

— 32:18

Different engines ingest different fields, which is a distribution problem rather than a writing one. One engine reads the video transcript; the others see only descriptions. Publishing the full script as text and syndicating it puts the words where the rest can reach them.

34:24 - Every episode is an event, so every episode is an entity

Every episode is an event, so it is an entity, so with enough social noise it can earn its own knowledge-graph ID and knowledge panel — not just the show.

"every single episode is an event. And if you think of it being an event, that means that it's an entity."

— 34:46

The inference is the interesting part. An episode is an event, an event is an entity, and an entity can earn its own identifier and panel. That means a catalogue can accrue entity authority episode by episode rather than only at the brand level — with enough social noise behind it.

35:37 - Uncontrolled pages get force-indexed and the owned site is left alone

Force-index the things you do not control — press releases, social posts, syndicated pages — but not your own site, because you want to see how your own site ranks naturally.

"To all these people that say indexing indexing is not important, it's not for your own website."

— 35:37

The asymmetry is deliberate. Pages you do not control may never be crawled at all, so they get forced. Your own site is left alone, because its organic crawl is a signal you want to keep readable rather than contaminate.

36:47 - When an answer fans a query out, the strongest entities get cited

When an AI answer fans a query out into candidates, the ones with the strongest entities get cited — so this off-site work influencing the models is more than on-site SEO.

"entities in the AI overviews and AI mode and Gemini the ones with the strongest entities seem to be the ones that get in."

— 37:05

When an answer fans a query out into candidates, relevance gets everything to the shortlist and entity strength decides who makes the answer. That puts the deciding work off-site, which is why it is described as more than on-page optimisation.

37:46 - Wire services differ, and a legacy news classification buys an extra crawler

Press-release wires vary; legacy news-classified wires (registered before 2019) get an extra crawler and extra rankings, and certain wire pages can trigger a top-stories carousel.

"we tested 27 different press release providers. Um we do that every quarter."

— 37:46

Wires are not interchangeable. A legacy classification earns an extra crawler and extra ranking surfaces, and certain pages can trigger a top-stories carousel. Testing them quarterly rather than trusting a list is the habit worth copying.

40:54 - A relabelled transcript manufactures attributed triples the models quote back

Relabel the transcript so each line reads "[name] says…", repeating the entity and forming a semantic triple that the models cite back with attribution.

"you're uploading a transcript f file."

— 40:54

The relabelling manufactures the format the models quote. Each line prefixed with the speaker's name repeats the entity and produces an attributed statement — exactly the shape that comes back with a citation attached rather than paraphrased into anonymity.

44:08 - Volume earns the standing to ask other businesses to publish consistently

Volume has a purpose beyond ranking: doing an extreme number yourself earns the authority to ask the businesses you back to do a modest amount.

"you turn up every single day doing videos and I'm doing exactly the same thing."

— 44:08

It removes the objection before it is made. Having produced at an extreme volume, asking a partner business for one episode a week is a modest request rather than an imposition. The volume buys standing, which is a return nobody counts.

44:22 - The system runs on a team and a refusal to let perfection block shipping

The system runs on a team and a bias to ship: hours a day filming, an offshore chain of editors, uploaders and thumbnail-makers, and a refusal to let perfection block progress.

"doing four to five hours a day doing videos every single day."

— 44:20

The system is a team and a bias. Four to five hours a day on camera, an offshore chain of editors, uploaders and thumbnail-makers, and thumbnails that are admittedly not as good as a specialist's — shipped anyway. Delegating everything but the filming is what makes the volume possible.

47:34 - Guesting beats self-hosting, because a third party supplies the description

Guesting on other people's podcasts is more valuable than only hosting your own, because you borrow their audience and gain third-party mentions of who you are.

"are there any platforms that you use that uh to to get on podcasts as a guest?"

— 47:34

Borrowing an audience is the smaller half. The larger half is that a third party introduces you, describes what you do, and the description carries weight your own site cannot give itself. Other people's words are the asset.

51:00 - A syndication engine can be sold directly as paid placement

A syndication engine can be monetised directly: inbound requests to appear can be charged as paid placements, and AI-visibility brands with big budgets will pay for the reach.

"monetization of a podcast."

— 51:00

Owning distribution others want makes it sellable. Inbound requests to appear become paid placements, priced per episode or as a package, and the brands with visibility budgets will pay for the reach. It is named as something that would not work on a pure audience show.

51:16 - Clip tools miss a niche, because they do not know what lands in it

AI clip tools are poor for a niche podcast because they do not know what lands in the niche or the audience's specific pain; picking the best clip needs someone who understands the subject.

"I tried them and they're all garbage. I like I I tried them because they don't because the thing is they don't know what hits in the niche."

— 51:16

The failure is domain judgement rather than capability. A model does not know what lands in a narrow niche or which pain the audience actually has, so the clip it picks is competent and wrong. Choosing the moment requires understanding the subject.

55:17 - The model is clonable: pick one topic and build the engine around it

The whole model is clonable: pick one specific topic, build the syndication engine around it, and once you have a following the relevant tool brands will come to sponsor you.

"if someone does decide to build a large podcast out,"

— 55:17

The recipe generalises. Pick one specific topic, build the syndication engine around it, accumulate a catalogue, and the tool brands in that space arrive with sponsorship. Nothing in the machine depends on the particular topic it was built for.

56:44 - Endorsement is only for tools already in use, because recommendations are trust

Endorse only tools you already use and would recommend anyway; testing products for pay costs time, forces you to share things you may not like, and risks the brand.

"the only times I get paid for recommendations are if it's a tool that I already use that I like."

— 56:44

A recommendation spends trust. Taking payment to test products costs time and forces you to share things you may not like, and the brand is what pays for it. Endorsing only what is already in use keeps the recommendation worth something.

1:00:30 - One standard guest-episode description turns every slot into an authority deposit

On guest episodes, use one standard description that associates the brand with its bottom-of-funnel keywords and links to a hub of landing pages — or to a specific underperforming page needing more link juice, with the language matched to the link.

"I just generally can jump on a podcast and go what do you want to talk about?"

— 1:00:30

Every guest slot becomes a deposit. One standard description associates the brand with its buying keywords and links to a hub of landing pages — or to a specific underperforming page, with the anchor language changed to match. Consistency is what accumulates.

1:02:14 - The biggest waste was the top of the funnel

The biggest wasted time was top-of-funnel keywords: there are far more bottom-of-funnel questions than expected, and going deep on a narrow one is "information gain" that ranks and gets shared.

"what are some of the things that you wasted time on that you wouldn't do again"

— 1:02:14

The regret is specific and worth more than the tactics. Top-of-funnel terms are competitive and convert poorly, and there turned out to be far more buying-intent questions than expected. Going deep on a narrow one is the information gain that ranks and gets shared.

1:03:55 - Almost every question-based term is easy to rank with video

With video, almost all question-based terms are easy to rank, so answering one specific question for five minutes gives a targeted bottom-of-funnel conversion.

"provide value and do bottom of the funnel to start with and then work the way up"

— 1:03:55

Few people make focused video for narrow questions, so the competition is thin and the intent is high. Five minutes answering one specific question is a complete asset. It is the whole model restated in one sentence, and it is where the episode stops.

Personal Branding Lessons

An episode about treating a recording as a ranking unit. The moves below are production decisions that follow from that one reframe.

Treat one search query as one episode

A tightly scoped asset matches a single intent and is easier to rank than a sprawling one, which is exactly the logic that governs a web page. Five to fifteen minutes, one topic, no wandering off subject. Six short episodes each aimed at one query beat one long conversation that touches all six, and the discipline transfers from page-building unchanged. 01:58

Target the bottom of the funnel rather than the how-to terms

A how-to query is somebody learning; a best-provider-for-a-situation query is somebody buying. Video ranks unusually well on the second kind, and the second kind converts. That single distinction is what redirects the whole catalogue away from explainers and toward the terms people type when they are ready to act. 03:40

Put the keyword in the title and the first line of the description

Front-loading is the trick, and it is the same on any platform: keyword in the title field, keyword in the opening line of the description, then a long bulleted description built from the transcript. Generated descriptions hallucinate, so the check is part of the step rather than an optional extra. 17:15

State the claim as a triple in the first sentence, then repeat it

Subject, verb, claim — a clean, well-formed statement said first and repeated throughout — is exactly the shape a language model can lift and attribute. The repetition is deliberate and close enough to keyword stuffing that it gets named as such. Frequency of one clean claim across ingested text is read as agreement rather than as one opinion. 18:06

Force-index the pages you do not control, never your own site

Uncontrolled pages — syndicated copies, social posts, wire releases — may never be crawled at all, so they get pushed into the index deliberately. Your own site is left alone, because its organic crawl rate is a signal worth keeping readable. The asymmetry is the whole rule. 35:37

Cut the backstory and give one line of credibility

Off-topic introduction is dilution measured against the searcher, who arrived for the answer. One line of credential supplies the trust, and everything after it belongs to the topic. The backstory serves the guest; the relevance serves the person who typed the query. 14:03

Guest on other people's shows rather than only hosting your own

Borrowing an audience is the smaller half of it. The larger half is that a third party introduces you and describes what you do, and a description in somebody else's voice carries weight your own pages cannot give themselves. Self-hosting at volume is a machine; guesting is the part that builds the entity. 47:34

Endorse only the tools already in use

Testing products for payment costs time, forces you to share things you may not like, and puts the brand behind them. Recommendations are trust, and trading them for a fee spends the thing the whole operation runs on. Taking an affiliate arrangement for something already used and already recommended costs nothing. 56:44

Questions

Each answer ends at the moment in the recording where it is given.

Can you use podcasts purely for SEO instead of for an audience?

Yes, and the two are treated as opposite jobs. An audience podcast is built on energy and daily return; a search podcast is built as a stack of keyword-specific episodes, each scoped to one query and made to rank once and convert. Nobody subscribes and binges it; someone finds a single episode at the moment of intent and turns into an inquiry. The volume — thousands of episodes in a year — exists because each one is a ranking unit, not an instalment. Deciding which of the two you are actually running is the first platform and format choice to make on purpose. 00:00

How do you structure a podcast episode to rank for a search query?

Build it like a web page. Scope it to one query — five to fifteen minutes, single topic, no fluff or backstory — then use the people-also-ask questions for that query as the questions you actually ask on camera. The test offered is: would this query justify opening a new page? If yes, the H2 questions that page would carry become the episode's structure. Supply just enough credibility in one line to earn trust, answer the questions concretely, and close on a soft call to action rather than a hard pitch. 01:58

What is the difference between top-of-funnel and bottom-of-funnel video?

A top-of-funnel term is someone learning — "how to do X" — and it is competitive, hard to rank for, and rarely converts. A bottom-of-funnel term is someone ready to act — "best provider of X for Y", or a competitor's name — and it captures a buyer at the moment of decision. The episode's whole model targets the second, and names the first as the biggest waste of time in a year of production. Video ranks especially well for buyer-intent phrasing, so the payoff for aiming low in the funnel is high. 03:40

How do you syndicate a podcast episode for backlinks?

Upload the recording, push it through a podcast host out to dozens of platforms, then classify each platform by whether it gives a do-follow link, a no-follow link, or strips links entirely. Force-index the useful ones so a crawler is made to visit them, because uncontrolled pages may otherwise never be crawled. Republish on a media database and build a guest profile whose "official websites" field carries do-follow links to the pages you want to strengthen. The result is many indexed, crawled, citable copies of one recording. 06:04

How do you get a podcast episode to feed and get cited by AI answer engines?

State the exact claim you want repeated as the title and the first sentence, then restate it throughout the transcript in a fixed subject-says-claim form, close to keyword stuffing, so the models lift it as though several sources agree. Publish the full transcript on the podcast page and syndicate it widely, because different engines read different fields — one reads the video transcript directly, others see only the description. Distribution across many authoritative copies is what builds the consensus signal answer engines weight, and what makes content spread through a system rather than a feed. 04:15

Does uploading the same video to multiple channels help or hurt?

Both, depending on your goal. Duplicating one recording across channels with different titles and thumbnails splits user signals, so each channel's click-through and algorithmic reach suffer — it hurts distribution. But it can trigger a video carousel in search, where several near-identical videos capture most of the slots a lone video cannot. So it buys placement at the cost of reach. It is offered as partly a test of which engagement signals matter, not a blanket recommendation; one channel plus wide syndication is all that is strictly needed. 24:08

What is the number-one ranking factor for video?

High audience-retention rate — how much of a video people actually watch — is named the top factor, bar none, and specifically for the ranking job, not just for being pushed. Channel quality and engagement are described as feeding the recommendation side, but retention is singled out as the one that outranks the rest. 26:18

How do you target a competitor's brand name with video?

A rival's brand name is itself a buying query, because people comparing options search it directly — and video ranks in a carousel just below the competitor's own site. Make comparison assets: "competitor alternatives", "competitor versus you". The honest version, which still captures the carousel, is to speak well of the rival and tell searchers to try both and keep whichever earns the better return. That even-handedness occupies the competitor's search real estate while reading as guidance, and it draws enquiries from people already in a decision — a move that rewards understanding what makes people trust one voice over another. 29:26

Should you use an AI avatar or AI clip tools for a ranking podcast?

For a specialist niche, the verdict is not yet. A cloned avatar and voice currently take the team longer than filming and sound robotic on technical terms, and the ranking value lives in the transcript and syndication anyway — a black screen with the right text still ranks. AI clip tools are judged poor because they cannot recognise what lands in a niche or which moment speaks to that audience's pain; picking the clip needs someone who understands the subject. Both judgements are marked as possibly dated as the tools improve. 20:37

Sources

He Did 2,000 Podcasts in a Year Just for SEO Rankings

The sections above follow the episode's own order. Each timestamp is the point in the recording where the idea is discussed, so the post reads as a map of the conversation rather than a rearrangement of it.

Quotations come from the episode transcript with spoken filler ("okay", "right", "you know", "um", "actually", "kind of", "sort of") and false starts removed. No words are added: square brackets mark anything inserted for sense, and an ellipsis marks any cut. British spellings are restored, and speech-to-text errors were repaired against context only where the intended word is unambiguous. Where a line appears both in a cold-open teaser and in its place in the conversation, the timestamp cites the second — the point at which it is actually said.

Quotes are stamped with the moment they are said rather than with a speaker's name.