AI Search Is Rewriting Marketing Funnel—and Podcasts Are Winning It

But the real story is knowing HOW podcasts and TV clips are fueling the new funnel success.

A recent stat shared that 49% of Mother's Day shoppers are now using AI to search for a Mother's Day gift. (As a mom, I'm wondering what these prompts are.)

This Sunday, NRF is forecasting a record $35.7B in spend, and Optimove just reported that 64% of those shoppers are open to buying from a brand they have never tried. That's not the biggest news.

This looks like a story about a holiday, but it's not. This is a story about mastering the new world of content.

How the expert CMOs and marketers are winning with AI (in a way you might not have considered)

We all know by now that AI insight is built from all of our collective content, and as my friend Ross Simmonds says, this is content put on the internet (not on your website) 10 years ago. The most innovative brands CMOs might not have a 10-year runway, but they are not buying more banner ads or sending more emails. They are doing something most of marketing has not caught up to yet.

The most innovative brands are feeding the model the content only they can make.

Banner ads do not train language models. Trade press releases do not train language models. Generic blog posts and landing pages get flattened into the same vector space as every other generic blog post and website, and the agent never surfaces them.

You know what does train models? Long-form audio and video from trusted third-party sources that gets transcribed. Yep. Podcast appearances (not your own podcast). TV interviews (including the sponsored ones). And sometimes the right conference talks that go up on YouTube with closed captions, plus Substack pieces with a clear date and a real name on the byline.

That is the content that ends up in the training set, in the search index, and in the agent's response when your future buyer asks who they should trust.

Of course reach is the byproduct, but the asset is the trained association between you and the topic.

As a TV and podcast host and a huge believer that PR, content, and brand are the secret behind the revenue machine for B2B and B2C, I have watched this happen in real time. The same point I made on stage at Davos shows up six months later inside an AI summary that is not pulling from my LinkedIn at all. It is pulling from the transcribed audio. The transcript is the asset. If you are not investing in these assets, today is the perfect time to start.

What the Liftoff Trust Index Deep Dives keep proving

At Liftoff Enterprises, my team and I run our proprietary Trust Index Deep Dive on real brands every week. The framework scores Authenticity, Inspiration, and Relatability on a 0 to 100 scale. The pattern is the same one Mother's Day just exposed at retail.

Case in point: we recently analyzed a Fortune 500 marketing services partner whose parent company works with 92% of the Fortune 500. They scored a 51, Trust Gap. The capability was massive. The AI search visibility was nearly zero. When we tested the queries their buyers actually use, the model surfaced their competitors instead. The infrastructure existed. The discoverability did not.

Another: a behavioral science platform with HBR, Forbes, and Netflix coverage in the founders' personal bios scored a 65, Trust Gap. The reason on the report read "invisible in every channel where CPOs and CHROs are searching." Gallup, with 26 million assessments behind it, owned every query. The brand owned none.

And the most expensive one: a North American product distributor sitting on an exclusive supply contract worth nine figures scored a 34, Trust Deficit. The product was real. The exclusive was real. The AI search engines confirmed they were invisible. Buyers asking about their exact product category were routed to competitors and offshore knockoffs. The exclusive was already worth more than the brand.

Same story, every time. The companies that score the highest on our Trust Index are not the biggest. They are the ones whose executives are on the right podcasts, the right TV shows, the right stages, and inside the transcripts the model reads.

Want to watch your revenue hockey-stick for 2H? Three moves to make.

Sponsor an interview on a trusted podcast. You don't need the biggest audience because it is not about distribution anymore (although that helps). It is about the podcast that knows how to get its transcripts into the market and onto YouTube clips. That is the one feeding the models. If a podcast publishes full transcripts, it is doing your brand more long-term work than a one-day press hit.
Invest in one piece of long-form content and a few short clips of it, from insights you actually own. A 20-minute solo episode, a sit-down video, or a webinar with a real transcript. Title it like a question your buyer would type into ChatGPT. "What does brand trust look like in the AI era" beats "Q2 thoughts on marketing." Models index questions.
Audit your top three appearances from the last 12 months. Are they transcribed? Are they dated? Is your name spelled the same way in all three places? Models reward consistency. Sloppy attribution is a tax you are paying every time a buyer asks the agent who knows what.

Bottom line: the funnel is not what you think it is anymore

For decades marketing taught us that the funnel went top to bottom. Awareness on top, consideration in the middle, decision at the bottom. We bought media to widen the top and we built CRM to optimize the bottom.

In 2026 the funnel is not a triangle. It is a feedback loop.

The new top of funnel is the LLM. The new middle is the agent that takes your buyer's prompt and goes shopping. The new bottom is a human who only sees three options because the model already filtered the other forty.

Your brand never reaches the human if the model does not know who you are.

Which means the question every CMO should be writing on the whiteboard this week is not "how do we get more reach." It is "what content is the model reading about us, and is it actually ours?"