The B2B Podcast Index
Index
All categories
MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
MethodologySubmit
Best of:MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
An independent project byFame
SearchBest episodesGuestsInsightsMethodologySubmit a podcast
Index/Marketing/Marketing Over Coffee Marketing Podcast
Marketing Over Coffee Marketing Podcast artwork

GEO Kills the Listicle

Marketing Over Coffee Marketing Podcast · 2026-07-02

0:00--:--

This episode examines two major shifts in AI and content strategy. First, Penn breaks down Anthropic's latest Claude updates, particularly Sonnet 5's transition toward agentic workflows rather than chat-based interaction. He explains the 30% increase in token usage and the emerging strategy of delegating work to specialized models - using expensive Claude for planning and cheaper alternatives like Gemma4 or Minimax for execution, much like managing a team. The conversation then pivots to a critical discovery about listicle content strategy: Penn's testing reveals that once you mention more than five competitors in a listicle, you're actually harming your own SEO authority. Under old algorithms like BM25 and TF-IDF, listicles worked because duplicated content was discarded; modern LLM-based systems amplify the most-mentioned competitors regardless of ranking, meaning competitors at the bottom of every listicle end up gaining citation authority. This represents a complete inversion of traditional SEO strategy, driven by how large language models process text at scale. John Wall connects this to broader brand trends around personality-driven marketing, where advertisers now seek individual creators rather than channels, supported by tools like GEO (geographic SEO) that leverage domain authority and content libraries.

Key takeaways

  • →Claude Sonnet 5 is an agent-first model requiring 30% more tokens, making the pricing gap wider between Sonnet and Haiku - creating opportunity for alternative models like Minimax or Gemma4 in agentic workflows.
  • →Listicles listing more than five competitors actively harm your SEO authority in the LLM era because repeated competitor mentions get amplified by language models as authority signals, inverting the traditional SEO benefit.
  • →The shift from chat-based to agent-based AI means structuring workflows like team delegation: hand planning work to expensive smart models and execution tasks to lightweight models, avoiding unnecessary cost.
  • →Advertisers are now buying personality and domain authority (like Marketing Over Coffee's 20-year reputation) rather than channels, making GEO and content library depth critical competitive advantages.
  • →Meta's smart glasses add paywalls to features and enable undisclosed facial recognition for model training, exemplifying privacy-hostile design that undermines consumer adoption.

Topics in this episode

Agentic AIAnthropicClaude Sonnet 5Gemma4MinimaxHaikuBM25 algorithmTF-IDFLLM SEOGEONewmediabloggingSEM

Questions this episode answers

Why do listicles harm SEO now when they used to help?

Under old BM25 and TF-IDF algorithms, listicles worked because duplicate competitor mentions were filtered out; modern LLMs amplify the most-repeated competitor names as authority signals regardless of ranking position, so competitors listed in every listicle gain unintended citation authority.

What's the difference between Claude Sonnet 5 and the previous version?

Sonnet 5 is 30% more token-intensive than Sonnet 4.6 and is optimized as an agent-first model that makes frequent tool calls and external function calls, shifting the pricing strategy and opening space for competing models like Minimax or Gemma4 as workhorse alternatives.

How should companies structure AI workflows to control costs?

Assign planning and complex reasoning to expensive models like Claude Opus, then hand execution tasks (web search, code generation, content creation) to cheaper lightweight models like Gemma4 or Minimax, similar to delegating work to team members by skill and cost.

Why are advertisers moving away from channel-based to personality-based deals?

Search ads are no longer delivering ROI, so brands are investing in personalities with proven authority and content libraries - like John Wall with Marketing Over Coffee's 20-year domain reputation - because that accumulated trust drives better results than channel placement alone.

What privacy issues does Meta's new glasses have?

Meta enabled facial recognition without announcement to train models on users' data and added a paywall to the Conversation Focus privacy feature, raising concerns about undisclosed data collection and every feature becoming subscription-gated.

Conversation analysis

Computed from the transcript - who did the talking, and the words that came up most.

Share of words spoken

  • Speaker C52%
  • Speaker A46%
  • Speaker B1%

Most-used words

marketing13model12interesting12framer11glasses10coffee9plan8listicles8models7competitors7site7cool6point6claude6back6sonnet6

Episode notes

In this Marketing Over Coffee: Learn why this is the end of the listicle, Fable is back, tire inflators win, and more!! Direct Link to File #3 in Malawi New Claude - Fable is back! Opus 4.8, Haiku 4.5, Sonnet 5 Using Gemma 4 for deep research to save on tokens The World Cup's Great […] The post GEO Kills the Listicle appeared first on Marketing Over Coffee Marketing Podcast.

Full transcript

Transcribed and scored by The B2B Podcast Index.

Speaker A: Foreign.

Speaker B: This is Marketing Over Coffee with Christopher Penn and John Wall.

Speaker C: Good morning. Welcome to Marketing Over Coffee. I'm John Wall. I'm Christopher Penn and we are the number three business podcast in Malawi, Africa. So I just had to throw that out there because that's just so cool that people in another part of the world are listening to us. There's no other point to that headline but that. So in more useful news, uh, Claude has been updated. We have become all Claude all the time. So what's the latest coming out of them?

Speaker A: Uh, so the new model which dropped yesterday, uh, which would have been June 30 as on the day of recording, it's funny, it actually dropped while I was in the middle of a client call talking about Claude and it's like, oh, update. Like, oh, all right, well so that's changed everything. Um, so there's, there's four models in the Claude family now. There's the fifth generation Fable, which we talked about previously, which apparently just got reapproved for use by the U.S. department of Commerce. So that will be back shortly. There is Opus, which is at version 4.8, which is their fourth generation, very smart model, very expensive. There is Haiku version 4.5 which is their super lightweight, basically documentation reading model. And there is Claude Sonnet which is sort of their mid range model. Sonnet just got upgraded to Sonnet 5. It's their, this is the fifth generation model, the second in that family besides Fable. And I've been testing it out for the last uh, 24 hours. Um, and notice a couple things. One, it is more token intensive than its predecessor, about 1.3. So about 30% more token intensive than Sonnet 4.6 which is the previous version and it is agentically tuned. So basically if you look at this, this is uh, clearly a derivative of their other 5th gen models. Just made a more practical workhorse. It's somewhere between Sonnet and Opus in its quality. Its speed is slower and boy does it make a lot of tool calls. So it uses a lot of external functions. Anthropic has said this is because it is an agent first model. So which is an interesting thing because it really shows that the industry overall is pivoting away from chat based models where you're having a conversation to agent based models where you hand over a project plan and say, okay, you do it.

Speaker C: Yeah, well, and that matches with all the Salesforce connection stuff. You know, we've been running too, as far as them just going all in on everything agentic all the time, like you said stepping further away from the chat. Yeah. How about as far as just the increased token usage? I mean, is that actually going to change the way you're looking at things or it's more just a matter of like, okay, we just have to work around this.

Speaker A: If you are used to using these tools in an agentic fashion, it's not going to change a lot because you're still following the basic recipe of use the smartest model to build the plan and have the lighter weight model execute the plan. What's interesting is that because Sonnet is more expensive, but Haiku, which is their lowest model, is still pretty dumb, there's now this weird little gap in between where uh, if you wanted something that was a good workhorse but not too expensive, it no longer squarely sits in that place anymore. Now on par with sort of just below Opus 4.8, which opens the door to using third party models. You know, outside of those, you know, if you are a Claude shop, you might want to look at something like either a GLM 5.2 or a minimax M M3. I've been using Minimax a ton and you could even use a local model like Gemma4 or Quin 3.6 to do the workhorse stuff. Right. Copy this HTML, write this code based on this plan. Search the web for this. So like, I use um, Gemma4 for deep research now because in an agent framework I can say, here's your 12 page research plan. Just follow the plan. And it's like, okay, I'll go Google this, I'll go Google this. And you know, it comes back an hour and a half later saying I did all the Googling and like, good job, you googled. And now I can hand it back to the smart model for processing. So one of the things that people be, need to be really thinking about is how you structure your workflow with these tools and how you decide who gets what kind of work from the different AI agents. In a lot of ways it's no different than delegating to different team members on your team to say like, yeah, you're going to do this and you'll do that and you'll do that.

Speaker C: Yeah, right. Isn't it? It's like managing a human staff as far as like, okay, these are the cheaper laborers so they can do the Googling and the actual creation of the plan goes to the more expensive model and having to move that stuff around. That makes a lot of sense.

Speaker A: Uh, except nobody calls in sick, right?

Speaker C: Yeah, right. You can have stuff run overnight and yeah, there's never any, never any human drama to go along with any of that. Ah, speaking of human drama, the International World cup, all kinds of stuff going on with that. The Great American sleepover on TikTok had in app searches surging 525%. Uh, it's just been interesting to watch kind of how this all went. I mean, there was a lot of concern in the months leading up to this as far as how this would all go and what's gonna happen, but it seems like it's done fairly well. And FIFA, of course, m minting one bajillion dollars out of all this is basically the end punchline. Yeah, I don't know if anybody has interesting stories on that. We've been pitched by a bunch of people that have not proven that they're worth, uh, letting them take your time. But, um, that's on there. The other one on the watch list has been creators at Con, the sessions going on. And I had a few creators that were, you know, writing about what's going on there. And, well, I mean, they were writing about themselves, of course, and surprisingly that was the big story. But there were a number of statements talking about how the big brands want to meet with the social media powerhouses. You know, that that's really where they're thinking the money is and the kind of stuff that they want to go into. It'll be interesting to see kind of where this goes. Yeah, if you're an up and coming creator, there's definitely more opportunity than ever right now.

Speaker A: Well, you've been talking about that at the office and stuff, about how advertisers are really looking instead of at a channel, they're looking at a, a, uh, personality. So instead of saying like, we want to be all in on email marketing, we're going to hit these lists, we want to be all in on these podcasts. You've been hearing that people saying, I want to work with John Wall specifically, I don't care on what channel. Like, what's the deal on that?

Speaker C: Yeah, well, I think one huge thing, you know, for me and the stuff that we've done is I think GEO is paying off on that. Right. Because we've got the marketing over coffee domain that has over 20 years of reputation and we just have this huge library of content that tracks with stuff because, yeah, I've had people kind of come out of the wood that, um, you know, our new voices and faces. But yeah, marketing over coffee lines up well with that. And so, I don't know, it seems to be the same kind of thing across the board. People are realizing that, hey, having search ads is not doing it for us at all anymore. That's not, you know, we were throwing money at that and you're not getting the same kind of response. Uh, so. Yeah, and it's interesting too that this does lead into you because you had a story talking about listicles. Like, it's interesting how that has changed completely too. Like there used to be value for people in listing their competitors and everybody in the space, but that's, that doesn't pay off the same way now because of the way these large language models work. Right?

Speaker A: Yeah. So I did a test to see how Listicles were treated by AI at scale. And so this, uh, this is based on some work done by Lily Ray, uh, the SEO expert out of New York City. She was saying she, she couldn't disambiguate between ubiquity versus authority and look, but in her empirical test show, companies were actually experiencing decreases in authority, uh, with Listicles. And so I set up this experiment to test the ubiquity portion. And it turns out from a language, uh, processing perspective, once you get above five competitors, uh, including you doing listicles in a content space, you're doing more harm than good to yourself. Because yes, you're at the top of your listicle, but a like 85% of listicle is devoted to your competitors. You're automatically giving them airtime. And then once enough people start doing it in a space, people don't always list the same competitors. So the competitors who get mentioned the most benefit, even if they're not doing listicles, because, you know, this, that and the other companies, all, you know that competitors, this, you know, this company is number five. Is number five in every listicle. Right. They're at the bottom of the heap. But because they're in every listicle from a text mass perspective, they're going to be the answer that gets cited instead of the people who are publishing the listicles hoping that AI will interpret them as being the number one company for whatever. So it is an interesting experiment to do in terms of how the machines process data. And it does show it's a tragedy of the commons. Once enough people start jumping on the bandwagon, the bandwagon goes off the rails real fast. And that was a really mixed up analogy.

Speaker C: Yeah. But, uh, it really hits me because for so long that was a big SEO thing of like doing these lists of your competitors was actually stealing their traffic. And now it's the other way around if you're writing this kind of stuff, you're actually going to get burned.

Speaker A: And it worked under old SEO because there were algorithms like BM25 best match 25 which would see that same, you know, low ranking competitor across everything and assume okay, let's discard this because it's the same across all these documents. Uh, there's also a variant called uh, TF IDF inverse document for format where if something is common across a bunch of documents, toss it out because it's commonplace and what you're looking for is unique. So in old SEO listicles work in AEO geo listicles are harmful because you're giving more mass to a competitor to somebody else.

Speaker C: Okay, well and as far as site management too, we have to mention Framer and thank them for their support of marketing over coffee. When a new landing page turns into a pile of tickets and handoffs, Framer helps your team move faster. Teams can collaborate in real time, iterate on the same page and publish instantly. Agents help you go beyond the vibe coded site and a first draft to bring you a production ready site faster than ever. It's a complete website platform, not just a builder, so teams can launch and keep improving their sites in one place. Thousands of businesses from early stage startups to Fortune 5000 uh, are choosing to build their websites in Framer, where changes take minutes instead of days. Agents and Framer work alongside teams to streamline collaboration on the same canvas, build custom code components, create, manage CMS content, optimize SEO settings and ship everything all in one place. Framer is an enterprise level solution powered by agents that provide premium hosting, enterprise grade security and 99.99% uptime service level agreements that give the world's leading brands like Perplexity and Miro the confidence they need to build their websites in Framer. Learn how you can get more out of your site from a Framer specialist or get started building for free today at ah.framer.com MOC for 30% off a Framer Pro annual plan. That's framer.com MOC for 30 percent off F R A M M E R.com MOC rules and restrictions may apply. Yes, take all the headache out of your life. As far as site management. Again, framer.commoc and we thank them for their support of the show. Okay, Meta glasses. Uh, I forget even there was another vendor that had some glasses about a month ago that looked ridiculous and social media immediately dogpiled on that and beat the hell out of them for their ridiculous looking glasses. Snapchat was the Snapchat glasses.

Speaker A: It was the new Snapchat glasses which are like goggles.

Speaker C: Yeah, right. Uh, the articles were all along the lines of like this is what happens when you're on a commit committee that's afraid to say no to the boss. You know, like he's telling them that these look cool. But Meta firing back now with glasses by Kylie Jenner going for more fashion. Meta seems to have been leading the charge of this and I've had a few people say that they love them. But uh, you were talking about the privacy notice for these things. How horrible is it? What's going on?

Speaker A: Well, so there's a bunch of things wrong with Meta's glasses, one of which is they've added a paywall to the Conversation Focus feature. And a lot of folks are saying, well that you're just going to get a paywall for every feature on these things now. But the bigger issue was they in an announced change to the software enabled the facial recognition feature. So the uh, Meta glasses were collecting facial recognition data from people's um, use of these things and using it to train models. Uh, is what the various news stories have ah, uh, cited. So it is. I mean this is par for the course for Meta. They're not exactly known as being either an ethical or a good company.

Speaker C: But you know, yeah, yeah, this is just part of the thing. I don't, I still just don't see any compelling reason and you know, to add insult to injury is the, I'd have to get prescription versions for me to really wear them. And so now you're talking about thousand dollar glasses and even um, sliding that through my health savings account is not enough for me to jump into that. So yeah, I don't know, I guess these things are going to sit on the background right, With VR headsets as kind of like fringe things that nobody really wants. Um, so we'll see how that goes and where that ends up.

Speaker A: I get the idea of wearing the glasses and stuff like that and having the camera is cool. If the device was under my control, I miss like the old Google Glass system was actually a pretty intelligent design. It looked dumb. I mean it looks like a Borg, but you know, the idea that you could flip up and down the display in the camera was a clear indicator A, that you're wearing some kind of wearable and B, when it's in the opposition that you ain't recording. So like you could tangibly show people, yeah, you're not on camera because it's pointing at the Ceiling.

Speaker C: Right, Right. This is. It's safe for us to talk and do things in this point right here. Yeah. And I'm still waiting for, you know, the whole, um, visual display of, you know, when I look at people, it reminds me what their name are and stuff like that. You know, things that actually could be useful to me, which, uh, are not there yet. So we had some other stuff. I had Gear Watch, an interesting thing. Um, my chair, I was looking at a new headrest. There's actually Atlas headrest, has speakers mounted in the headrest. So for calls, you could actually have these Bluetooth speakers just behind your head so you don't have to wear earbuds. And it's one of those things where I was, you know, the gear nerd in me was like, oh, my God, I need to buy these instantly. And as the more I thought about it, I'm like, well, will I really use that? Probably not. And of course, the big one for me is I always fall back on. Like, I hate having to charge stuff. You know, having to, like, I don't want to have a cord on my chair. That's just a recipe for disaster. And having to, you know, of course it will die on the most important phone call of my life. Like, it won't die. You know, when I'm getting some kind of spam call, it'll be at the worst possible time. So that was kind of cool, but I ended up passing on that. But I'll keep an eye on that.

Speaker A: I got the new anker Soundcore Liberty 5 Pro headphones because I wanted to try out the $169 Vers. They were not on sale on prime day. Um, I'm impressed. Like, the earpieces themselves, you know, they're very thin silicone, so those probably will get changed out at some point, maybe just foam chips. But the noise cancellation on them is pretty decent for the price point. Better than other things at the price point where it really shines. I was. I was genuinely shocked. Was on the microphone. The microphone sounds like a, uh, V Moda. Like, it sounds good. Very, very, very good. Which for Bluetooth headphones is kind of nuts. And what Anchor was publishing was that they have the. Basically, they're doing the same thing Adobe Podcast does. They're reconstructing your voice as opposed to trying to isolate it from the environment. Like, okay, that's a pretty smart design.

Speaker C: Yeah, right. It's interesting to hear talking about that because that was kind of the whole next level of, like, having all this onboard hardware that can just remix Your whole voice, autocorrect, auto tune, like you know, and then the noise is gone because it's just a, uh, representation of you. And yeah, as long as the words match, like nobody cares if it actually sounds right. And so you did though you skipped the more expensive version where subscribe headphones. I know that that's uh. You just passed on that feature.

Speaker A: Yeah, they have the AI transcription feature which still requires a cloud. And I'm like, you know what, I don't need that. I've already got my own hardware and stuff to do at home. I don't need to be paying yet another $20 a month to somebody for a feature that I literally have at home. So I passed on that there are people who are going to use that and they're going and stuff like that. For me, I'm like now, ah, there's, there's no reason for it.

Speaker C: Yeah, the endless pile of subscriptions is, you know, annoying in itself. That's a whole other thing. Actually talking about subscription free, I had deep discounts.com if you're into movies and music and physical media stuff. They're a great site that has some discounts. Um, and as part of that too, I actually watched Sinners over the weekend which I'm so behind on. It won a few Oscars for last year, but it was actually, if you're into vampire movies, it's actually worth checking out. So I wanted to throw that out there. Um, and then I had the Chorus 3 running watch which I've been doing a lot of testing with. That's actually proven itself to be pretty good. And the even more interesting one was a Highline tire inflator I had. I got a flat on my bike last week. And yeah, you have this USB charged inflator that is if you've ever been one of those knuckleheads on the side of the road with the hand pump trying to get your bicycle wheel up to 70 pounds. That is not a thing anymore. Go buy one of these things. You know, the good ones cost 80 bucks or whatever, but it's just complete game changer of, you know, 15 seconds and you're done and you're back on the road as opposed to sweating for 40 minutes to try and get that done. So, um, definitely throwing that out there for life changing tech for anybody that needs that. Let's see, how about anything else have you got as far as travel coming up or other stuff that's you're working on?

Speaker A: Uh, we just signed for some events in September. I also have Some online stuff this month I'm doing a thing with Marketing Prospect, uh, marketing profs, uh, I think as weekly webinars that I'm doing on um, agentic AI and things. And they're already uh, asking folks about, you know, to make sure that people know about marketing. Across B2B forum this November, I'm going to be doing a completely brand new session that doesn't exist yet, uh, on agentic B2B marketing. So it should be interesting to see, you know, how can you build a marketing team that is solely machines? And the answer is yes. And how well it works yet to be seen. But, uh, that's what's on the docket. But yeah, a lot of stuff coming up this fall. It's going to be a very busy fall.

Speaker C: Cool. And then I had some other content just I wanted to plug. One was Robert Riggs, who is a friend of Trust Insights. He has a podcast about drone warfare. And so that has just blown my mind in the past month. This idea, you know, for all of history that I've studied, people are like, you never get to Moscow. Like trying to get to Moscow is what kills every empire. And Ukraine is throwing drones right into downtown Moscow, which is completely nuts. So I don't want to get trapped down that rat hole. But if you want to go down that rat hole, I'll have links to Robert's podcast. And the other one is AI Psychosis. This idea of people kind of just going far afield and losing their way because of too much AI. Um, Chris, you and Katie covered that in the Trust Insights podcast. So I'll have a link to that. But any other, you know, teaser you want to throw on that, as far as what you guys covered there, it's a real thing.

Speaker A: The psychology community is still figuring it out in a lot of ways. Like there's no DSM 5 entry for it, but they are definitely consequences for it because it, it's about these tools and their psychop. Fancy reinforcing things that they shouldn't. Like a human friend go, dude, that's stupid. Don't do that. You know, whereas a machine like, that's a great. You're so smart, John. Yeah, that's. So your actual friends are like, no, dude, that's boneheaded. Uh, don't do that. And that, and that's kind of where these tools are at now is that they are at a point where they, they agree too readily and they have no understanding of context. They don't recognize the warning signs in aggregate going, you know what it might be time to step away from the computer for a little while.

Speaker C: Right, right. That, that's on there. Okay. So if people want to dig more on that, we'll have the link, uh, to the Show Notes over there. You can sign up for the Text line at 617-812-5494 or the newsletter over on the site. But otherwise, uh, yeah, stay cool. Don't, uh, get heat exhaustion. That's going to wrap it up for us. So until next week, enjoy the coffee.

Speaker A: Enjoy the coffee.

Speaker B: You've been listening to Marketing over, uh, coffee. Christopher Penn blogs@christopherspenn.com Read more from John J. Wall at, uh, jw5150.com the marketing over Coffee theme song is called Mellow G by Funk Masters. And you can find find it at Musically from Mevio or follow the link in our Show Notes.

Related episodes across the Index

Other episodes covering the same guests and topics, from across The B2B Podcast Index.

  • Eric Ries on Why Good Companies Go BadPodcast Archives · on Anthropic92 / 100
  • Decision Logic: The Difference Between an Answer and a DecisionThe AI Forecast · on Agentic AI87 / 100
  • 512. Is SpaceX Over or Undervalued, Why Consensus Kills, How Chewy Beat Amazon, and the GameStop Saga from a Board Member (Larry Cheng)The Full Ratchet (TFR) · on Anthropic86 / 100
  • KYA Won't Always Protect You. The Real Risk Is the Swarm!Fintech Conversations & Insights with Efi Pylarinou · on Agentic AI86 / 100
  • Agentic AI in Sales: What Business Leaders Need to KnowScaling with AI · on Agentic AI86 / 100
  • DeepSeek's $50B Round, OpenAI's Delayed IPO, and the GP Stakes Market with CAZ Investmentstrading places · on Anthropic86 / 100

More from Marketing Over Coffee Marketing Podcast

All episodes →
  • Proving the Advantage that Marketing Creates75 / 100
  • Salesforce CMO Amber Armstrong at Connections 2026!
  • Now with More Synthetic Performers and Less Fable!
  • UChicago Medicine CMO Andrew Chang at Salesforce Connections!
  • Katie Robbert on Putting Claude Cowork to Work
Explore the best B2B Marketing podcasts →
All Marketing Over Coffee Marketing Podcast episodes →