The Ten Blue Links Are Going Away. Start a Podcast Anyway — Even If Nobody Subscribes.

For twenty years, content marketing had one shape. Write the article. Rank for the query. Collect the click.

That shape is coming apart. AI Overviews now answer a large share of questions directly on the results page. The user gets what they came for and does not scroll. Ranking fourth for a question that Google has already answered above you is not the asset it was in 2021.

The usual conclusion is that this is a catastrophe for content marketing. We think that is the wrong read — and that the right response for most businesses is one they will not expect.

Start a podcast. Even if nobody ever subscribes to it.

Not for the audience. For what falls out of it.

Discovery didn’t die. It moved — and Google told us where.

This is the part that is usually argued from vibes. It does not have to be.

On 7 July 2026, Google shipped a new property type in Search Console: platform properties. You connect your Instagram, TikTok, X or YouTube account, and Google reports how that account’s content performs in Google Search. It went globally available on 29 July. No website required.

Sit with what that means. Google built, staffed, shipped and globally rolled out measurement infrastructure for social and video accounts inside its search tool. Companies do not do that for surfaces they consider marginal. You do not build a reporting product for something that does not matter.

That is Google stating, in the most concrete way available to it, that your video and social content is part of your search visibility — not a separate marketing activity happening somewhere off to the side.

So the honest summary of 2026 is not “search is dying.” It is: the text-article-to-blue-link path is compressing, and Google is simultaneously formalising video and social as a discovery surface it measures.

The clicks did not vanish. The format that earns them changed.

Why a podcast is the cheapest way to produce for that surface

Here is the argument that actually decides this, and it is an operational one rather than a philosophical one.

Sit down for one 45-minute recorded conversation. Count what you can legitimately produce from it:

• A long-form video for YouTube
• An audio episode for the podcast apps
• A transcript — thousands of words of indexable, quotable, machine-readable text on your topic, in your named experts’ own words
• Eight to twelve vertical clips for Instagram, TikTok and Shorts
• Quotes and stills for static posts
• A newsletter section
• Usually, a blog post, if you want one

Now count what a written blog post produces. One blog post.

That ratio is the entire case. A podcast is not really a show — it is the most efficient content-production format available to a business, because a recorded conversation is the only raw material that converts into every surface at once. Everything else requires you to make each asset separately.

And the surface Google just started measuring — short video on social — is the one a podcast produces in volume, almost as a byproduct.

“Even if nobody subscribes” is not a rhetorical flourish

Most businesses that abandon a podcast do so for the same reason: they looked at download numbers after four months, saw a few hundred, and concluded it was not working.

They were measuring the wrong thing.

If your show produces ten clips a month, and those clips are being surfaced in search and social to people who have never heard of you, the podcast is doing its job whether or not a single person subscribes to the feed. Downloads measure the audience for the format. The clips are the distribution, and their reach is not in that number at all.

This reframe matters because it changes what you optimise. A show run for its clips is booked, recorded and structured differently from a show run for subscribers. You care about whether each episode contains six genuinely quotable ninety-second moments. You stop worrying about episode-to-episode continuity that only serial listeners would notice.

It also lowers the bar to starting, which is the real barrier. You are not committing to building a media property. You are committing to a recurring recorded conversation that feeds everything else.

The thing a podcast produces that AI cannot

There is a second reason, and it is going to matter more over the next two years, not less.

The internet is filling up with competent generated text. Anyone can produce a serviceable 1,500-word article on any topic in about ninety seconds. The floor has risen and the median has flattened, and search systems are increasingly having to work out which sources actually know something.

A recorded conversation is expensive to fake in the way that matters. It has a named person, a face, a voice, and a specific opinion they are willing to attach to themselves. It contains the things that only come from having actually done the work — the caveat, the war story, the “well, that depends, and here’s when it doesn’t.”

That is first-hand experience in the most verifiable format available. It is the raw material that generated content cannot manufacture, and it is what your best clips will be made of.

What this looks like practically

If you are starting from nothing, the version that works is smaller than you think:

Record every two weeks, not every week. Twenty-six episodes a year is plenty of raw material, and a fortnightly cadence survives a busy quarter. Weekly shows are where good intentions go to die.

Record video from the start. Audio-only forecloses most of what makes this worth doing. You do not need a studio — a decent camera, decent light and a decent microphone, in that order of increasing importance.

Have a real expert on the mic. Someone in your business who genuinely knows the subject. The whole advantage here is authenticity that cannot be generated, so putting a non-expert host in front of a script gives up the one thing you had.

Book guests your customers would want to hear from. A guest brings their own audience and their own credibility. It is also, quietly, the best business development channel most companies never use — a 45-minute recorded conversation is a warmer first meeting than any sales call.

Plan for clips at the recording, not after. Ask questions that produce self-contained answers. When a guest says something excellent, let it land — do not talk over the end of it. A clip needs a clean beginning and a clean end, and you can create both while recording, or spend an hour hunting for them afterwards.

Publish the transcript. It costs nothing and it is thousands of words of substantive text on your topic. Skipping it is throwing away most of the search value.

What we are not claiming

We are not going to tell you that clips get you into AI Overviews. Nobody outside Google can promise that, and the people selling you a guarantee are ahead of the evidence.

We are not claiming podcasts are free. They are a real, recurring commitment, and a show that stops after five episodes is worse than one that never started.

And we are not claiming written content is dead. Well-researched, genuinely useful writing is doing fine. What is struggling is the middle — the competent, unremarkable article written to rank, which is precisely the category AI both answers directly and produces infinitely.

What we are claiming is narrower and, we think, hard to argue with: discovery is moving toward video and social, Google has now formally begun measuring it, and a recorded conversation is the cheapest way to produce for that surface at volume.

The businesses that will be visible in 2028 are the ones with a large library of real people saying specific things on the record. That library takes years to build, and the only way to have one then is to start now.

The MPN Awards are open to every marketing podcast — membership is not required. If you are thinking about starting a show, see why podcasters join the network.