Best of LinkedIn: AI in B2B Marketing CW 43/ 44

00:00:00: This episode is provided by Thomas Allgaier in Freeness, based on the most relevant LinkedIn posts about AI and BDB marketing.

00:00:07: In calendar weeks thirty-three and thirty four.

00:00:10: Freenes is a bdb market research company that supports enterprise marketing teams in unlocking full potential of their customer data with help from AI.

00:00:19: So imagine logging into your computer tomorrow morning And forty pieces highly researched SEO content all your social video clips for the week and Your entire outbound email queue have just been completed.

00:00:36: Not just drafted right now

00:00:37: fully finished, right?

00:00:39: Finished mm-hmm And it all happened while you slept.

00:00:43: This isn't some theoretical pitch for twenty thirty.

00:00:47: It's a reality that unfolded Just last week

00:00:50: I mean fundamentally rewrites what a marketing department even is.

00:00:54: We are watching the entire foundation of this industry shift in real time,

00:00:59: we really are and you know B to be.

00:01:01: marketing is getting hit by a massive wave on two fronts simultaneously because internally our day-to-day execution is moving away from human heavy lifting toward autonomous AI agents but externally Our buyers or entirely abandoning traditional search engines in favor of AI generated answers.

00:01:19: Yeah, the landscape is totally unrecognizable.

00:01:22: Totally!

00:01:23: So today for this deep dive we are breaking down the exact mechanics of how you can navigate this new reality.

00:01:30: We should probably start with that internal shift because The future of B-to-B Is unequivocally AI driven.

00:01:37: But the first place You as a marketer Are going to feel That impact As right inside your own daily workflows.

00:01:43: We are moving miles past the era of using an AI as like a glorified spellchecker.

00:01:49: Now it's being deployed as an autonomous worker.

00:01:52: Yeah, that overnight automation scenario I just mentioned is a perfect illustration of this.

00:01:57: There was this brilliant experiment posted by Eric Sue That breaks down what this looks like in practice.

00:02:03: His team actually set up six specific grokbot agents Designed to complete and entire days worth marketing tasks.

00:02:10: Six bots

00:02:11: yeah.

00:02:12: So like an SEO bot hunted down forty viable pages to create content for, while a short-form bot systematically cut video clips and an outreach bot built cues of warm emails.

00:02:24: That's wild!

00:02:24: But you know the crucial detail there is how those bots actually operate.

00:02:29: they aren't just running a single rigid script

00:02:32: right?

00:02:33: They are multi agent systems.

00:02:35: one bot finishes a task evaluates the output And then hands it off By the morning, the human marketers just log in.

00:02:43: Review is centralized approval list and you know hit go on a full day's work.

00:02:47: Well

00:02:47: wait let me stuffy there because setting up complex hive of six bots talking to each other sounds like an absolute nightmare for our standard marketing team.

00:02:57: Oh sure

00:02:58: Are we all supposed to become developers overnight?

00:03:01: No not at all.

00:03:02: And the thing is, you don't need a massive complex bot ecosystem to see these gains.

00:03:07: The mechanics of automation have become incredibly accessible

00:03:10: which actually brings up A completely different approach To this that Ira Bodnar mapped out because instead Of building autonomous bots he built forty eight reusable clawed skills.

00:03:22: All right

00:03:22: He essentially created a library of specific tasks.

00:03:25: for example He has a skill That automatically flags wasted Google ad spend like anything over twenty dollars with zero conversions.

00:03:32: That is so useful,

00:03:33: right?

00:03:34: Or a skill that spots meta-ads create a fatigue before the cost per acquisition starts climbing.

00:03:40: Yeah and the mechanism behind those skills?

00:03:42: what makes them so powerful?

00:03:44: when we say he built a skill What actually did was to create highly specific system prompts And context files.

00:03:50: Okay break it down for

00:03:51: me.

00:03:51: So imagine handing a freelance marketer A massive binder of your Brand guidelines, historical data and strict rules before they ever look at your ad account.

00:04:00: That is basically what a skill file does.

00:04:03: it grounds the LLM in your specific business logic Before processes any new data.

00:04:08: Oh I see

00:04:08: Yeah you set that file up once And from then on You just feed It The raw export and ask it to run the analysis.

00:04:14: Well

00:04:15: Actually i saw another hyper-specific use case for that kind of Data structuring From Anthony Carlton.

00:04:21: He took a thousand unformatted CRM contacts.

00:04:24: Oh, we all have those messy lists?

00:04:26: Yeah just the first name and a random email.

00:04:29: but he used Claude to enrich the entire list.

00:04:32: in ten minutes

00:04:33: Just ten minutes

00:04:34: Ten Minutes.

00:04:34: The AI figured out the company...the industry And then segmented every single contact by which of his specific consulting offers fit them best.

00:04:44: I mean..he completely replaced an eighty dollar-a month software tool just By using an LLM that categorizes existing data.

00:04:50: And data structuring is precisely where marketing teams lose the most hours.

00:04:55: Gabe Rogel and Ryan Vong highlighted some recent G-II research that puts this friction into perspective.

00:05:01: What did they say?

00:05:03: Well, fifty six percent of marketers cite cross team handoffs as their absolute top challenge while only twenty one percent blame their actual tech stack.

00:05:11: Oh wow So the software works fine.

00:05:14: It's just a human friction slowing everything down

00:05:16: Exactly.

00:05:17: The bottleneck always happens when work moves from strategy to execution or from one department to another.

00:05:23: And that is why campaign builds are the ideal.

00:05:26: first AI workflow you should automate

00:05:28: because it removes their handoff,

00:05:29: right?

00:05:30: The underlying technology allows you to automate the routing of campaign briefs directly into task managers like Asana or marketing platforms like Marketo.

00:05:39: The AI handles all the repetitive administrative translation between teams and you only insert human agents where you know, nuanced commercial judgment is required.

00:05:48: Okay let's unpack this because this is the part of the automation conversation that always makes me a little nervous.

00:05:54: The quality

00:05:54: control

00:05:55: Exactly If we put everything on autopilot...the writing..the data routing....the ad analysis aren't we just building high-speed factory for garbage?

00:06:04: I mean Is there risk to our brand identity here if we just let machines run assembly line?

00:06:09: It is very real threat.

00:06:11: Philip Edzel coined a phenomenal term for it Good Slop

00:06:14: Good slop.

00:06:15: That's hilarious,

00:06:16: but it's true.

00:06:17: the danger in twenty-twenty six isn't that AI will make terrible Unusable content.

00:06:21: The models have gotten incredibly good at mimicking your brand colors Your fonts.

00:06:25: you're general corporate tone But they fundamentally lack what he calls?

00:06:30: The taste layer.

00:06:32: ah that uncanny valley feeling where a graphic looks Technically correct.

00:06:36: at a glance something just feels completely hollow.

00:06:38: exactly The taste layer is honed by ten thousand hours of human expertise.

00:06:43: It's the subtle spacing, graphic treatments and nuanced tone that a human designer just feels intuitively.

00:06:50: Right!

00:06:50: The machine doesn't have taste.

00:06:51: Exactly...the real danger is your C-suite or sales teams now have access to these generation tools right on their phones.

00:06:59: If they use uncurated AI to generate assets for flyer presentations Your brand gets diluted across the market by a thousand tiny, slightly off variations.

00:07:11: So how do you actually fix that structurally?

00:07:13: The fix is shifting oversight from reviewing individual graphics to managing system level.

00:07:19: You have build heavily curated strictly guided AI skills.

00:07:23: your whole company has mandated use.

00:07:25: That makes total sense.

00:07:27: Real quick before we pivot to the other half of this equation.

00:07:30: How our buyers are utilizing these exact same tools?

00:07:33: Just a reminder for you listening, if your finding this breakdown helpful and want us stay ahead on those trends make sure to hit subscribe so that you don't miss out future deep dives!

00:07:41: We track these shifts as they happen.

00:07:46: If internal marketing teams are deploying agents to execute campaigns, it stands to reason that our buyers are deploying their own agents to research those campaigns.

00:07:55: Right?

00:07:55: We're fundamentally moving from a B-to-B model... ...To A-B-A Model.

00:08:01: Business To Agent.

00:08:03: I love the framing so much!

00:08:04: It changes the entire paradigm of who we were even talking about.

00:08:09: Manpreet Wada and Eli Schwartz brought this b-toa concept into the forefront recently.

00:08:14: AI agents are now actively shortlisting vendors before a human buyer ever even visits a corporate website.

00:08:20: Wow!

00:08:21: And to grasp the scale of this shift, look at Anthropic.

00:08:25: they recently released their three hundred and thirty-five page S one IPO filing okay?

00:08:30: The word SEO or search engine does not appear as single time in that massive document.

00:08:36: Wait...not

00:08:36: once

00:08:37: Zero.

00:08:38: In over three hundred pages detailing your future with internet

00:08:40: zero mentions.

00:08:42: Anthropic doesn't view search engines as their competition because traditional Search is rapidly becoming a background API tool.

00:08:49: Just data feed for the AI?

00:08:51: Exactly!

00:08:52: The old model, you know...a user typing keyword scrolling ten blue links and clicking website.

00:08:58: that's being replaced by agent loop.

00:09:01: How does it work?

00:09:02: An autonomous agent takes an open-ended business problem retrieves background data from those hidden search APIs evaluates options against its system prompt and then produces a finished recommendation for the buyer.

00:09:14: The human never even sees the

00:09:15: links.".

00:09:16: That is wild!

00:09:17: And Holly Chan at Indrayan actually provided some hard numbers backing up how fast this was happening.

00:09:22: Fifty-one percent of software purchases now start with AI search...

00:09:27: More than half?

00:09:28: ...more than half, but there's massive disconnect in how marketing teams are measuring that because we're still obsessing over short traditional keyword prompts like ranking four CRM Software.

00:09:41: But real buyers using agents don't search like that.

00:09:43: They give the agent long complex questions with highly specific constraints built in, Like team size budget integration requirements

00:09:52: because The AI system can process those long tail constraints Instantly to filter the entire market

00:09:58: exactly.

00:10:00: And there is a canyon of difference between an AI simply mentioning your brand in a bulleted list versus actively recommending Your brand as the absolute best fit for a buyer.

00:10:10: specific constraints.

00:10:11: Yes, because if you're just mentioned You are just noise.

00:10:14: but if you recommended you were actually In the pipeline

00:10:18: exactly.

00:10:19: But here's where it gets really interesting.

00:10:20: from me B to A sounds a bit like trying to get past the digital bouncer at really exclusive club.

00:10:26: That's great way of putting it, right?

00:10:28: If the bouncer, The AI agent doesn't recognize you your sales team does not even get inside the door and make a pitch.

00:10:34: Yep

00:10:35: But the million dollar question is Is this actually translating into real revenue today or is just theoretical visibility metric for future?

00:10:43: Oh It is absolutely translating in Revenue Right now.

00:10:46: Marcus Sheridan and Guy Tano Nino DiNardi highlighted this beautifully.

00:10:50: One marketer received three massive speaking leads in a single week from prospects who explicitly stated, we found you on chat GPT.

00:10:57: Wow!

00:10:58: Yeah the pipeline is tangible.

00:11:00: however there is a crucial caveat here that marketers need to understand about generative engine optimization or GEO.

00:11:07: Okay what it?

00:11:08: Geo does not create new demand.

00:11:11: It only captures existing in-market demand.

00:11:14: Let's pause on that because that feels like huge distinction.

00:11:20: If a buyer asks an LLM for the best product analytic software, they already have a problem.

00:11:27: They are actively in market.

00:11:29: The LLm is just doing the sorting Right.

00:11:31: So if your overall software category is shrinking or people don't fundamentally know you need your type of solution.

00:11:37: yet Ranking high and chat GPT won't magically invent new buyers.

00:11:42: GEO ensures that you win the buyers who're looking.

00:11:45: Okay so...If AI agent is the bouncer And we want to capture that existing demand.

00:11:52: What exactly are the criteria for getting on the VIP list?

00:11:55: How are these LLMs actually choosing which brands to cite and recommend over

00:11:59: others?".

00:12:00: This is where we get into the pure mechanics of how these models retrieve information.

00:12:04: Ben Wills in Maya Voya analyzed research across one hundred different industries, to find the number-one ranking factor for LLM's.

00:12:12: The strongest signal by far is relevant industry backlinks and brand mentions And notably, those mentions do not even need to be hyperlinked.

00:12:19: Wait

00:12:20: really?

00:12:21: So just the raw text of your brand name existing in proximity To relevant industry topics is enough.

00:12:26: Yep

00:12:27: How does an LLM Even weigh a plain Text mentioned without link?

00:12:31: That goes against everything we know about SEO

00:12:33: Because LLMs don't click links The way Google's traditional crawler Does.

00:12:37: They read vector embeddings

00:12:39: Okay what does that mean?

00:12:39: In plain English

00:12:40: They map the semantic relationship between words in a massive multi-dimensional space.

00:12:45: So if your brand name constantly appears and paragraphs discussing B to be market research, The model mathematically links your brand To that concept.

00:12:54: Oh I get it.

00:12:55: interestingly A ref status shows a massive divergence Between traditional search an AI search.

00:13:00: because of this only thirty seven point one percent Of the pages cited in Google's AI overviews actually rank In the top ten of traditional search for That exact same query?

00:13:10: And if you're wondering about traditional technical SEO tactics, they found that adding schema markup barely moved the needle on AI citations at all.

00:13:17: Wow!

00:13:18: You know there is actually a highly actionable way to trigger these trust signals on your own site.

00:13:24: Ross Hudgens and Alex Groberman ran a test and found that mid-authority brands that place five or more credibility statements directly on their homepage see a sixty two percent lift in AI visibility.

00:13:37: Sixty

00:13:38: Two Percent When they say credibility statements, what specific markers are the models actually looking for?

00:13:45: They're parsing for entity trust signals.

00:13:47: Basically things that prove you are a legitimate functioning business.

00:13:51: so review counts exact usage volume years in business.

00:13:55: reputable customer logos

00:13:57: make sense.

00:13:58: right because The AI doesn't have human intuition So it relies heavily on these proxy signals for Trust.

00:14:05: Furthermore if you want to optimize for Google's discovery engine specifically They found it heavily favors content that is chunked into small sections of fewer than three hundred and seventy-five words structured with clear question based headaches.

00:14:18: Oh

00:14:18: interesting

00:14:19: Yeah, it apparently really struggles to extract the answers from massive, unbroken walls of text.

00:14:24: That actually aligns perfectly with how a retrieval augmented generation system or RAG parses and pulls information.

00:14:31: but this raises an important complication regarding platform diversity which was brought up by Will Leatherman, Melissa Rosenthal, Al Lettolani ,and Michelle J Raymond.

00:14:39: okay every single AI engine behaves drastically differently, you just cannot treat them as a monolith.

00:14:45: They really are completely different beasts

00:14:47: right.

00:14:47: for instance chat GBT sites about three sources per answer on average but perplexity which is built entirely for research cites about nineteen.

00:14:55: well

00:14:55: that's huge gap yeah

00:14:57: and Gemini relies heavily on YouTube videos for its answers while ChatGBT almost completely ignores youtube in it citations.

00:15:04: And what's crazy?

00:15:05: Is the data source feeding these engines or constantly shifting underneath our feet.

00:15:10: like ChatGPT's reliance on Reddit just plummeted by eighty-eight percent recently because Reddit implemented strict login walls to block the scrapers.

00:15:18: Right, they

00:15:19: cut them off?

00:15:19: But at that exact same time LinkedIn's share of chat GPT citations jumped a staggering four hundred and thirty nine percent in a single month.

00:15:28: That is massive.

00:15:30: So tracking a single blended AI visibility score across all these platforms Is basically trying to win a decathlon.

00:15:37: only practicing high jump Right?

00:15:39: Every platform plays by completely different

00:15:41: rules.

00:15:42: Yep

00:15:43: So does this mean we just have to endlessly spam our brand name everywhere on the internet and hope that vector embeddings pick it up?

00:15:50: No, definitely not.

00:15:52: breadth is far more important than repetition.

00:15:55: Glenn Gabe an Alex Halliday looked at Cystrix data which monitors us And It shows wild daily fluctuations in Brand Mansions.

00:16:03: Why so volatile?

00:16:04: because These answers are dynamically generated based on temperature settings and server loads.

00:16:09: So an LLM might cite you on Tuesday, And then completely drop you on Wednesday for the exact same prompt.

00:16:16: because of that volatility covering a wide variety of different buyer prompts in highly specific use cases is vastly superior to trying to force your brand name to rank For one specific keyword over-and-over.

00:16:27: so cast a wider net to catch The dynamic variations

00:16:30: exactly And speed is critical right now as these models solidify their knowledge bases.

00:16:36: Brands that managed to appear early in AI answers over the summer captured thirty six point seven percent of total visibility by the fall, late newcomers so those exact same prompts they captured only three point six per cent.

00:16:48: That's a punishing drop off

00:16:50: It Is!

00:16:51: The AI short lists are forming and becoming mathematically very sticky.

00:16:56: So with all of this technology moving so incredibly fast, I mean we've got our internal agents executing our campaigns and buyer agents deciding which software makes the VIP shortlist.

00:17:07: What exactly is The Human B-to-B Marketer supposed to do in twenty twenty six?

00:17:12: Are We All Just Transitioning Into Bot Managers?

00:17:15: I mean, it is incredibly easy to feel that way looking at all the technical data.

00:17:20: But Voltaire Taub and Sinan EK really grounded this conversation in fundamental business logic.

00:17:27: Before you build any complex AI stack or worry about engineering cloud skills You have to clearly define your foundational strategy.

00:17:35: What is your unique value?

00:17:37: Who was your exact audience, what category are you actually playing in?

00:17:41: because an AI system fundamentally doesn't know the answers to those questions.

00:17:45: It only knows when you feed it

00:17:46: exactly.

00:17:47: AI can generate fifty variations of ad copy and three seconds but Only a human strategist with deep commercial judgment Can look at those fifty ads and decide which specific message The business Actually needs To deploy to solve its current growth bottlenecks.

00:18:01: that Is so true.

00:18:02: Actually, Gina Sapien and Morgan B dug into the current hiring data on this analyzing over eleven hundred marketing job posts.

00:18:11: The reality of the job market might surprise people who feel behind on the tech.

00:18:15: What did they find?

00:18:16: Over fifty percent start-up job descriptions don't even mention AI skills.

00:18:20: yet

00:18:20: Really?!

00:18:21: Yeah!

00:18:21: And only seven percent general marketing roles mentioned autonomous agents.

00:18:26: So Who is actually doing heavy technical lifting to build these b.toa pipelines?

00:18:30: Well, the complex technical AI builds are increasingly being handed off to dedicated GTM engineers and marketing ops teams.

00:18:37: Oh okay.

00:18:38: Yeah senior marketers are still being hired for things machines fundamentally cannot do commercial judgment strategic vision and deep customer empathy.

00:18:48: there's actually a great example of research regarding YouTube strategy.

00:18:52: an LLM can not replace human youtube strategy team.

00:18:55: why is that?

00:18:56: The reason it fascinating LLMs inherently regressed.

00:18:59: They operate by averaging out existing data to predict the most likely next word.

00:19:05: they cannot grasp the subtle jagged nuances of audience psychology Pacing or human humor that makes a video actually go viral.

00:19:14: That makes perfect sense.

00:19:16: So what does this all mean for you?

00:19:17: The listener.

00:19:19: it sounds like AI isn't replacing the marketer at all.

00:19:21: It's just becoming the ultimate sous chef.

00:19:24: Oh, I love that right

00:19:25: it chops the onions lightning speed if perhaps the ingredients flawlessly.

00:19:30: But you are still the head chef who has to look at all those ingredients, understand what the critics want and design.

00:19:35: The actual menu.

00:19:37: that analogy hits the nail on the head And Thomas Ross highlighted exactly why?

00:19:42: That Head Chef role is so vital to business survival right now.

00:19:45: Gartner predicts that by twenty-twenty seven seventy percent of enterprise brands will report AEO answer engine optimization metrics directly to their executive leadership.

00:19:55: Wow

00:19:56: it would be a core board level KPI.

00:19:59: but Are teams actually executing on that yet?

00:20:02: Because I'm looking at these numbers and the gap is terrifying.

00:20:05: It's a massive gap!

00:20:07: Right now, only twenty percent of marketers have started doing this optimization work.

00:20:12: Meanwhile ninety-four percent of B to V buyers are already using AI to make their purchasing decisions.

00:20:18: Twenty percent of sellers adapting to ninety four percent of buyers.

00:20:22: That is market failure waiting for

00:20:24: it.

00:20:25: And the roadblock isn't a lack of technology.

00:20:27: The tools are cheap and available, as we've discussed.

00:20:30: The Roadblock is a lack leadership... ...a lack cohesive strategy.. ..and a lack cross-team ownership.

00:20:36: Companies are paralyzed in meetings wondering who owns AI visibility?

00:20:40: Is it the SEO team...?

00:20:42: Is it PR..?

00:20:43: Is it demand gen…?

00:20:44: While they argue over org charts….

00:20:46: …the AI agents already shortlisting their competitors?

00:20:49: We have covered a massive amount ground today from deploying overnight multi-agent bot teams to optimizing your home page for the digital bouncer of AI Search?

00:20:59: Yeah, and I want to leave you with one final thought to mull over as you audit your own strategies this week.

00:21:05: As we move deeper into twenty twenty seven.

00:21:08: if buyers are exclusively using Ai agents to evaluate and shortlist vendors And vendors are exclusively Using ai agents to market and reach out to buyers We are rapidly approaching a reality where two algorithms are essentially negotiating with each other in the dark.

00:21:23: Oh man!

00:21:24: Right, and a world were the machines primarily talk to the machines.

00:21:28: how will your human trust?

00:21:30: Your real-world reputation...your physical relationships....Your unfakable expertise become your ultimate untouchable competitive advantage?

00:21:40: That is a great question.

00:21:41: end on if you enjoyed this episode.

00:21:43: new episodes drop every two weeks.

00:21:45: Also, check out our other editions on field marketing channel and partner marketing account based marketing, more tech go to market.

00:21:52: And social selling.

00:21:53: Thank you so much for joining us For this deep dive.

00:21:56: Don't forget to subscribe.