Best of LinkedIn: AI in B2B Marketing CW 33/ 34

00:00:00: This episode is provided by Thomas Allgaier and Frennis based on the most relevant LinkedIn posts about AI in B to be marketing.

00:00:06: In calendar weeks, thirty three and thirty four.

00:00:09: Frennis is a B to b market research company that supports enterprise marketing teams in unlocking Imagine ranking number one on Google organically for years.

00:00:23: Oh, yeah the dream

00:00:24: right.

00:00:24: you're relying entirely On that pipeline and then you just completely vanish overnight Just because a buyer asks an AI the exact same question.

00:00:32: it is terrifying

00:00:33: It really is.

00:00:34: so we have this massive stack of curated insights from B to be marketing practitioners who are basically on the front lines Right now.

00:00:41: And The mission of today's deep dive Is to figure out the mechanics behind these massive shifts?

00:00:46: Yeah Because looking across all these sources This overarching thesis is pretty undeniable.

00:00:50: I mean, the center of gravity in BDB marketing has just moved entirely.

00:00:54: Oh for sure The industry's no longer debating.

00:00:56: you know whether to adopt these technologies like that.

00:00:58: conversation Is just over right?

00:01:00: The mandate now is operationalization

00:01:02: Exactly!

00:01:03: The real problem is how You actually integrate These generative systems without breaking your brand trust or well destroying Your domain reputation or

00:01:13: data architecture.

00:01:14: Right, compromising that fundamental data architecture.

00:01:16: So if the whole landscape is changing I feel like the most logical place to start as very top of the funnel Like how buyers even find you in this new environment.

00:01:26: Yeah it all starts there

00:01:28: And Bjorn Raddy points out a really fundamental pivot here.

00:01:31: Traditional SEO has always asked You know How do i rank?

00:01:35: But Generative Engine Optimization or GEO It asks completely different question Which Is... and AI trusts.

00:01:45: And the underlying mechanics of those two questions are just completely opposed, I mean traditional search engines or essentially librarians?

00:01:51: Yeah they're just fetch things.

00:01:52: Exactly!

00:01:53: They retrieve links based on keyword density in backlinks but AI answer engines their synthesizers

00:02:00: Right there actually reading material.

00:02:02: yeah they read it and construct an answer based upon what they perceive as a factual consensus.

00:02:08: Nicholas Ndall shared an RF's analysis that perfectly illustrates this disconnect.

00:02:13: Okay,

00:02:13: what did they find?

00:02:14: So it looked at millions of cited URLs in Google AI overviews and They found that only thirty eight percent Of the sighted sources were actually in google's traditional top ten search results.

00:02:27: Wow so That's a staggering disconnect.

00:02:29: It is!

00:02:30: It implies that the signals google uses to rank you know A standard blue link are heavily divorced from the signals its own AI uses to validate a definitive answer.

00:02:40: Precisely,

00:02:41: and Delphine Vazal will back this up with his own observations.

00:02:44: being a legacy brand or category leader with massive historical domain authority it does not give you free pass into AI generated answers

00:02:53: because they're looking for something else.

00:02:54: right generative models look for structured information in clear entity relationship.

00:02:58: so if your website is full of like marketing fluff rather than definitive logically structured statement.

00:03:04: Yeah,

00:03:05: I just ignore it

00:03:06: exactly.

00:03:06: It simply cannot parse your expertise.

00:03:08: man The most concerning part of this remarketing director has to be the volatility.

00:03:13: oh yeah

00:03:14: Thomas Ross actually noted that this visibility is incredibly unstable.

00:03:18: He tracked one brand that was cited ninety-four percent of the time for their own brand queries.

00:03:23: Wait, ninety four

00:03:24: percent?

00:03:24: Yeah, ninety-four percent.

00:03:25: and suddenly That dropped to nine percent For non-brand queries And they would just frequently vanish completely from day today!

00:03:33: That's

00:03:33: wild

00:03:34: Right.

00:03:34: he is telling teams that weekly monitoring Is now an absolute requirement.

00:03:39: To understand why it so volatile you have look at how Retrieval Augmented Generation or ARA Actually works under The hood.

00:03:46: Okay.

00:03:46: Unlike a static search index, when the user prompts an AI The system fetches dynamically generated set of context documents to build its answer.

00:03:55: Right.

00:03:56: So microscopic changes in users prompt or even slight adjustments to the AIs internal weighting It changes which document gets pulled into that context window

00:04:05: You could be top.

00:04:06: answer on Monday.

00:04:07: By Thursday model weights have shifted and you were entirely excluded from response generation

00:04:13: Which means the old SEO playbook like earning a top spot and checking a dashboard once a month is totally obsolete.

00:04:19: Completely

00:04:20: obsolete.

00:04:20: So how does a marketing team actually build a durable strategy for this?

00:04:23: Yep, then he should go.

00:04:24: par-gava argues that we have to stop obsessing over a single magical AI visibility score.

00:04:32: She notes, frameworks like SEMrush are now tracking roughly eighteen different KPIs prompt gaps, all of

00:04:46: it.

00:04:46: It requires a total architectural shift in how you build your digital presence.

00:04:50: Really

00:04:50: does.

00:04:51: Anna York shared a twelve point landing page checklist for earning the specific type of algorithmic trust and what really stands out is her emphasis on entity based language.

00:05:01: Yes Let's break that down for second because entity-based language kind of sound like a term engineers use just to confuse marketers?

00:05:06: I

00:05:07: assume this basically means moving away from stuffing long tail keywords instead.

00:05:12: You know writing map relationships.

00:05:15: That is exactly the mechanism and AI doesn't read a sentence like we do.

00:05:19: it maps nodes.

00:05:21: so if you are a cybersecurity company, The AI is looking for definitive statements that connect subject verb object structures, and she also emphasizes providing a real hero asset like proprietary data backed up with verifiable case studies.

00:05:45: The AI is scanning for evidence not persuasion.

00:05:48: That makes Anne Smarty's observation incredibly relevant here.

00:05:51: She pointed out that about pages are now mission-critical for generative AI citations, which is funny because historically the About page where companies just throw up some corporate headshots and a vague mission statement then they never update it again.

00:06:03: Right!

00:06:03: Just an afterthought.

00:06:04: But from perspective of an AI crawler like ChatGPT or Gemini... ...that About Page is the root node in your entity graph.

00:06:12: It

00:06:13: actively crawls those pages to assess legitimacy Like, who are the founders?

00:06:17: What is the verified history of the company.

00:06:19: what do your physical location?

00:06:20: so

00:06:20: it's vague.

00:06:21: if that page lacks structured factual data The AI just lowers this confidence score in your entire domain.

00:06:28: Wow Kyle Atwater Morley added another layer to this strategy.

00:06:32: I think solves a lot of the confusion.

00:06:35: AI systems demand A clear point-of-view.

00:06:39: yes they do

00:06:40: like.

00:06:41: broad topical coverage Is actually a liability.

00:06:43: now His example was brilliant.

00:06:45: He said, Great Jones is the Dutch oven you give for milestone

00:06:49: moments.".

00:06:49: That's so specific!

00:06:50: Right?

00:06:51: You have to plant a flag with one sentence angle prove it on your own site and then build off-site proof points across the web So that AI naturally associates your brand at that highly specific concept Because

00:07:01: if you don't define the category For The

00:07:02: A.I.,

00:07:03: the A.

00:07:03: I.'s

00:07:03: just going To Define It For You Yeah And will likely lump you in With ten other generic competitors

00:07:07: Exactly.

00:07:08: But there Is a massive risk In treating this purely as an algorithmic game.

00:07:12: Lies Adam shared a cautionary tale about this.

00:07:15: Over a period of just nine days, the share of chat GPT answers drawing on Reddit citations plummeted from thirteen percent to two point.

00:07:23: two percent

00:07:23: so meaning If a brand had just spent six months seeding subreddits to manipulate chat GPT's training data,

00:07:30: their entire pipeline evaporated in the little over

00:07:32: week.

00:07:32: That is brutal!

00:07:33: Yeah chasing algorithmic loopholes is incredibly fragile.

00:07:37: The point there is that living your brand being genuinely useful and creating durable first-party trust across every touchpoint Is the only strategy that actually holds up when open AI or Google inevitably changed their architectures

00:07:52: Absolutely Before we look at how marketers are attempting to circumvent this search volatility by using AI for outbound sales.

00:08:00: By the way, if you're finding this deep dive helpful For your strategy make sure to subscribe so you don't miss our future additions.

00:08:06: Yes definitely subscribe.

00:08:08: We know inbound search is volatile right now, and the natural reaction for a lot of B-to-B companies has been to buy fully automated AI sales bots.

00:08:16: To basically brute force their way into buyers inboxes.

00:08:20: Yeah!

00:08:20: The hype cycle over last year promised us AISDRs that never sleep

00:08:24: And the pitch was incredibly seductive Right?

00:08:26: Plugin an AI agent.

00:08:28: Give it your ideal customer profile.

00:08:29: It will magically handle prospecting personalization Timing judgment.

00:08:34: You just sit back watch meetings roll In

00:08:36: But reality completely different.

00:08:38: Mick Kidower shared this satirical, but honestly painfully accurate story.

00:08:43: He joked about replacing his human sales team with fourteen AISDRs.

00:08:48: Oh no!

00:08:49: Yeah, they sent nine thousand completely generic emails asking for a quick question that they never actually stated Of

00:08:55: course.

00:08:56: Then they generated forty-thousand personalized videos featuring weirdly blinking avatars That consistently mispronounced the prospects company names.

00:09:05: Oh...that is painful.

00:09:06: And the grand result of all this automation was four opens and one single reply.

00:09:10: Asking if he's alright.

00:09:12: That's hilarious.

00:09:13: But the satire hits hard because of the actual market data reflects it.

00:09:17: Jonathan MK shared some hard numbers that act as a massive wake-up call for industry.

00:09:22: What were those numbers?

00:09:23: The most heavily funded AISDR company, a platform called Eleven X reportedly lost seventy eight percent of its annual recurring revenue right past the ninety day break clause in their contracts.

00:09:36: I would argue that tells us everything about the mechanism of failure here.

00:09:40: Companies are signing up, getting dazzled by the demo and then within three months they realize the AI lacks the contextual awareness to manage a multi-step email thread.

00:09:49: Exactly!

00:09:50: It hallucinates details damages the domain reputation with spam complaints And customers exercise their opt out clauses immediately.

00:09:58: Didn't Artisan another massive AI SDR player just post job listing for human BDR?

00:10:04: They did.

00:10:05: Yeah, they absolutely did.

00:10:06: And when the company selling fully automated AI sales reps have to hire human sales reps to sell their product?

00:10:13: The market has spoken!

00:10:14: Yeah...the irony is thick.

00:10:15: Very.

00:10:16: The underlying data shows that AI-only SDRs generate fewer meetings trigger higher spam filter rates and produce drastically lower meeting to opportunity conversions than human hybrid models.

00:10:27: So if replacing a whole human is failing, we have to look at what actually works?

00:10:31: Josie Germain provided really good structural alternative here.

00:10:35: He pointed out that instead of buying a five thousand dollar-a month tool That tries to act like a human which as we saw sees a seventy percent churn rate The answer is narrow automation.

00:10:45: Narrow Automation yes.

00:10:47: he built a highly specific tool for one manual bottleneck which was answering complex security questionnaires.

00:10:53: That task used to take two days of waiting on a busy CTO provide answers.

00:10:58: Such kills momentum!

00:10:59: Exactly, now sales rep gets the CTO level technically accurate answer in

00:11:05: fifteen minutes and that distinction between broad autonomy and narrow automation is crucial.

00:11:11: Eugene Kadzin and Johnny Fransich, who actually builds go-to market systems for major enterprises.

00:11:17: They both stated plainly that autonomous AISDRs just do not work reliably with today's technology.

00:11:22: Yeah the tech isn't there yet

00:11:24: And There is a massive legal blind spot most vendors are ignoring too.

00:11:27: Eunice Emery Auskaya pointed out that the EU AI Act, specifically Article Fifty requires companies to explicitly disclose when an artificial agent is communicating with a person.

00:11:37: And almost nobody's building that disclosure into their outbound workflows?

00:11:41: Nope they're just pretending it's real-person.

00:11:43: It really feels like the tech industry is obsessed with building fully autonomous you know level five self driving car for sales.

00:11:52: All the sales reps actually need right now is cruise control and a highly accurate GPS system.

00:11:58: That's perfect analogy.

00:11:59: Give them map in speed but let humans steer car

00:12:02: And that maps perfectly to brilliant historical parallel drawn on sources.

00:12:06: regarding medical field Back in twenty sixteen, Jeffrey Hinton one of the absolute pioneers of deep learning, publicly seated that medical schools should just stop training radiologists because AI would completely replace them within five years.

00:12:20: Oh I'm familiar with this.

00:12:21: it's a classic example of The Jebens Paradox.

00:12:24: AI made interpreting a medical image drastically cheaper and faster.

00:12:28: so because the friction dropped hospitals simply ordered vastly more images...

00:12:31: ...the demand skyrocketed

00:12:33: Right!

00:12:33: And Because the volume of raw images exploded the human judgment required to finalize those diagnoses them to the patient, that became The Premium bottleneck.

00:12:43: Exactly and the exact same dynamic is happening in B-to-B outbound.

00:12:47: Craig Rosenberg and Oshin Vowell both emphasize this in the sources.

00:12:51: How so?

00:12:52: Well AI makes drafting and sending a hyper personalized message practically free.

00:12:57: So every company increases their volume by ten X buyers get overwhelmed And inbox filters naturally escalate To block the noise.

00:13:04: Right it's an arms race.

00:13:06: Yeah who can exercise genuine empathy and strategic judgment, like knowing precisely why to call a specific buyer on a specific Tuesday.

00:13:18: That becomes infinitely more valuable

00:13:20: because you simply cannot automate empathy exactly?

00:13:23: You can't

00:13:24: man.

00:13:24: every marketer listening To this right now has likely had a CEO or board member ask them for their AI strategy This month And the temptation is just swipe up corporate credit card by few of these isolated tools to show activity in hope For The best.

00:13:36: oh I see it all.

00:13:38: But based on what we are seeing, how do we apply that premium human judgment at the structural

00:13:43: level?

00:13:43: Well.

00:13:43: That ad hoc approach is precisely why organizations are hitting a wall right now.

00:13:48: my JoJo outlined a fascinating five-level maturity model for go to market teams.

00:13:53: Okay What Are The Levels?

00:13:54: So level one is basically assisted AI.

00:13:56: you know your standard chatbots.

00:13:58: level two is automated where these tools run in isolation.

00:14:01: And she notes that almost the entire industry is currently stuck at level two,

00:14:05: so I'm guessing everyone stalls it.

00:14:06: Level Two because well connecting all of these isolated tools requires unglamorous data plumbing than nobody actually wants to pay for

00:14:14: right stuff.

00:14:14: Nobody Wants To Pay For Exactly.

00:14:16: i mean moving to level three requires actual structural engineering not just you know buying another software subscription.

00:14:24: You nailed It Because Level Three Is Orchestrated which means the tools actually share context and communicate with each other.

00:14:31: And getting there requires standardizing data schemas, building resilient API endpoints mapping out all those complex workflows.

00:14:40: Yeah that's a lot of heavy lifting

00:14:41: It is!

00:14:42: Most teams just stall-out because they refuse to do foundational work.

00:14:46: In fact Tony Larks noted that sixty percent of AI initiatives fail simply because underlying data isn't AI ready.

00:14:54: Because if you feed a fragmented siloed CRM into a large language model, the AI doesn't magically clean your data taxonomy.

00:15:03: Not at all!

00:15:03: It just scales your garbage data at machine speed across every single customer touch point

00:15:08: which is forcing very rapid evolution in build versus buy conversation.

00:15:13: John Miller shared an anecdote from dinner with go-to market leaders.

00:15:17: There's massive trend of marketing teams using AI to build their own internal software applications

00:15:22: which on paper sounds incredibly innovative and agile.

00:15:26: Oh sure, but the reality is an absolute nightmare for IT!

00:15:30: He mentioned his son VibeCoded, a CRM that got hacked in the single weekend.

00:15:34: Wow!

00:15:35: And for anyone unfamiliar, Vibe Coding is when you just prompt an AI to write software for without actually understanding underline code or security vulnerabilities it introduces?

00:15:44: Yeah...

00:15:44: It's risky.

00:15:45: When marketing teams build homegrown apps to bypass enterprise procurement they also bypass enterprise security compliance and maintenance.

00:15:52: Right

00:15:52: all of guardrails.

00:15:53: Exactly.

00:15:55: The consensus at dinner was building your own go-to market app creates an unsustainable technical debt.

00:16:02: Like who patches the zero-day vulnerabilities?

00:16:05: Who fixes the API when it randomly deprecates while one guy built is on

00:16:10: vacation?".

00:16:11: Yeah, this pushes us away from marketing tactics and into a much more serious conversation about enterprise infrastructure.

00:16:17: Caroline Bombart and Damon Burrell both highlighted that

00:16:21: shift.

00:16:21: Agentic AI is no longer a localized fun experiment for content team... It requires true enterprise architecture.

00:16:29: Boards of directors don't want to see a dashboard showing AI activity metrics, they demand AI-enabled business value that is secure and scalable

00:16:37: which inevitably brings up the topic of Enterprise Governance

00:16:40: Which I have to push back on here because whenever a BtoB marketer hears the word governance They instantly think of corporate red tape That just suffocates innovation before it can even launch?

00:16:50: It's very common anxiety But we need to reframe how we think about governance.

00:16:54: Okay has so

00:16:55: Governance isn't a red light at an intersection.

00:16:57: Think of it like a standardized electrical grid.

00:16:59: if your company doesn't have agreed-upon voltages and standardized plugs across the enterprise, people are going to start plugging their random two hundred forty volt AI tools into one hundred twenty volt wall sockets And they will literally burn the building down.

00:17:13: Oh I see

00:17:14: governance.

00:17:15: So proper data architecture security protocols and strategic alignment is simply making sure everyone Is using the same voltage?

00:17:23: I

00:17:25: love that framing.

00:17:26: So we've got the structural plumbing mapped out and we have this standardized electrical grid in place, but what are we actually pumping through these systems?

00:17:34: Because all of the infrastructure is completely useless if the actual output we hand to a buyer is terrible.

00:17:40: And it brings us into the core philosophical shift in modern brand strategy.

00:17:44: Marvin Cho introduced concept perfectly encapsulates this The Brand Bank Because generative tools exist, the actual execution of content like writing or designing and formatting is essentially free.

00:17:58: And the immediate instinct for many brands to flood every channel with high volume output.

00:18:03: They treat sheer throughput as a primary metric in marketing progress

00:18:07: Which is exactly that wrong approach.

00:18:09: When you pump out generic automated filler You are making quiet continuous withdrawal from consumer trust.

00:18:18: Consumers are already flooded with synthetic noise, which means the threshold for what constitutes good content has shifted entirely.

00:18:26: Adriana Preco shared a brilliant example of this.

00:18:28: What does she do?

00:18:29: She built an AI agent named Aria to draft her newsletter and The AI processed the inputs and generated a draft that was technically flawless.

00:18:38: It had clean grammar proper formatting perfectly optimized keywords

00:18:42: But I assume it felt completely hollowed to read

00:18:45: Lacked any real opinions, it took up space but said absolutely nothing of value.

00:18:49: Her takeaway is vital for marketers The competitive differentiators.

00:18:53: no longer whether an AI can write your content the differentiator's weather.

00:18:57: a highly skilled human validated what produced injected distinct perspective and insured actually meant something before was published

00:19:05: which ties perfectly into what Sarah Stella LaTanzio was talking about regarding the AI slot witch hunt.

00:19:11: Oh, The Witch Hunt?

00:19:12: Yes.

00:19:12: Platforms like Substack are currently rolling out A.I.

00:19:15: detection badges to publicly shame writers by showing what percentage of an article was machine generated.

00:19:22: but as she points out chasing percentages misses the point entirely.

00:19:26: knowing a diss is forty percent microwaved and sixty percent cooked by a chef doesn't tell you if the meal actually tastes good.

00:19:34: That is a great way to put it, and that's why she highlighted a company called Clay which has one of the best internal AI policies out there.

00:19:40: They don't ban AI usage.

00:19:42: they simply ask their team one question Can you personally stand behind every single sentence in this document?

00:19:48: Because mass AI content Is essentially empty calories.

00:19:52: It looks like full mail takes up real estate on screen But provides absolutely zero intellectual nutrition to your buyers.

00:19:58: Zero!

00:19:59: They consume it And leave feeling completely empty.

00:20:02: If execution is now a commodity, then editing tastes and strategic direction become the premium skills.

00:20:08: And if we follow that logic to its natural conclusion it fundamentally alters the organizational chart of a marketing department.

00:20:16: How so?

00:20:16: Well Rebecca Schultz and Rand Tamar pointed out That as AI successfully absorbs the execution layer of marketing organizations are going to flatten dramatically.

00:20:26: I would argue that AI's gonna brutally expose weak marketing judgment.

00:20:30: Yeah.

00:20:31: I mean, if an executive uses the efficiency of AI as an excuse to fire senior experienced marketers just to save a few dollars.

00:20:39: they aren't going get less marketing...

00:20:40: Oh!

00:20:40: They'll get more?

00:20:41: ...they are going to get exponentially more marketing but with absolutely zero strategic

00:20:46: direction.

00:20:47: Synthesizing everything we've unpacked today A clear picture emerges The winners in B-to-B Marketing over next three or five years will not be teams that have most AI subscriptions nor ones who produce highest volume of daily content.

00:21:00: The WITTERS will be the organizations that successfully architect-structured, governed and secure data foundations...and then combine those systems with deeply empathetic highly strategic human

00:21:11: judgment.".

00:21:12: Which builds perfectly on this brand bank concept from earlier…if execution is virtually free Throughput is no longer an impressive metric.

00:21:20: Discipline, your only real deposit into the brand bank?

00:21:23: Exactly!

00:21:24: So...the ultimate question for marketing leaders isn't how much you can produce.

00:21:28: The metric that truly matters now Is what you refuse to ship.

00:21:32: Like How many automated drafts are your team actively killing Every single week Before they ever leave building?

00:21:38: If you enjoyed this episode New episodes drop every two weeks.

00:21:42: Also check out our other editions on Field Marketing Channel and Partner Marketing Account-Based Marketing Martek, go-to market and social selling.

00:21:50: Thanks for joining us on this deep dive!

00:21:52: Don't forget to hit subscribe…and we'll see you next time!