Best of LinkedIn: AI in B2B Marketing CW 13/ 14

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

00:00:06: In calendar weeks, thirteen and fourteen.

00:00:09: Frennis is a b-to-b market research company that supports enterprise marketing teams and unlocking the full potential of their customer data with the help of AI.

00:00:17: you can find more info in the description.

00:00:20: Hey everyone!

00:00:20: And welcome to The Deep Dive.

00:00:22: today we're jumping into well...the absolute top AI and B to be marketing trends that we've been seeing across LinkedIn over the past couple of weeks.

00:00:31: So if you're navigating this space right now, You are definitely in the right place.

00:00:35: Yeah And just to set the stage for you listening We're giving ourselves a very specific mandate today Like, no fluff.

00:00:41: We are skipping the whole philosophical debate thing and the wild five years in the future thought

00:00:47: experiment.".

00:00:48: Right

00:00:48: nobody needs more of that right now.

00:00:49: Exactly our mission today is to take this massive just overwhelming volume of noise around AI And synthesize it into actual functioning operational models for modern marketing teams like what he's actually working And,

00:01:03: and maybe more importantly what's failing spectacularly right?

00:01:07: Which I mean.

00:01:08: that brings us directly to the first major theme.

00:01:10: we pulled from the content.

00:01:12: AI search is basically becoming a new front door for BDB buying.

00:01:16: Oh yeah The whole death of traditional SEO.

00:01:19: Totally!

00:01:20: I mean Marcus shared this stat.

00:01:22: That should honestly be massive wake up call if you're still running like A twenty-twenty three seo playbook.

00:01:28: Over last year he has been polling his live audiences just asking if they prefer AI search engines over legacy Google.

00:01:36: OK, and what were the numbers?

00:01:38: Well a year ago it was hovering around thirty-five percent Yeah.

00:01:41: But today across his audience is sixty five to seventy percent of people prefer AI Search.

00:01:46: Wow!

00:01:47: Seventy percent.

00:01:48: And that adoption curve only getting steeper.

00:01:50: I mean, when you look at why that shift is happening it makes complete sense from the perspective of a B-to-B buyer.

00:01:56: Like if i'm researching enterprise software... ...I don't want to dig through ten blue links leading to ten different vendor landing pages that are just bloated with SEO keywords.

00:02:06: Right exactly!

00:02:06: I

00:02:06: just wanted AI to synthesize the capabilities of all those vendors into one direct answer.

00:02:12: but that user preference It creates the structural earthquake for us as marketers.

00:02:17: Gillian Hart recently mapped this out as a transition from B to B, so business-to-business... To what she's calling A to B?

00:02:24: Business-to AI-to business.

00:02:25: Exactly!

00:02:26: The implications of that B to A to V model are just profound because I mean we've spent decades optimizing the top of the funnel to capture human attention.

00:02:35: yeah right but Hart argues that the funnel has now compressed entirely.

00:02:40: AI isn't just a research assistant anymore.

00:02:43: It's literally the new gatekeeper on the buying committee.

00:02:45: Yeah, the first entity that consumes your white paper is into human.

00:02:49: nope it's an algorithm and That algorithm parses the information if filters out all your marketing fluff And then it decides whether you're solution even makes it onto the shortlist that gets presented to The Human Decision Maker.

00:03:00: so wait

00:03:01: we're basically

00:03:01: tasked with optimizing our content So that machines can interpret it?

00:03:05: Just so they can then explain it to humans.

00:03:07: I was thinking about this earlier.

00:03:08: its kind of like Cooking a highly complex meal specifically so a food critic can digest it and then the critic just tells The diner how it tasted.

00:03:17: like you aren't feeding the diner anymore.

00:03:18: You're feeding the critic.

00:03:20: I mean, I see where you going with a food-critic analogy but If we look at the actual mechanics of these models It's really more like preparing a dish for a chemical analyzer okay?

00:03:29: A chemical analyzers.

00:03:30: that's a lot less romantic.

00:03:32: very much though because The AI isn't, you know, tasting your content or appreciating your clever copywriting.

00:03:38: It's parsing your schema markup and your entity relationships to see if your data satisfies the technical parameters of what the user is searching for.

00:03:47: So when Joseph and Scott Jones pointed this out explicitly...

00:03:50: Right they were talking about structuring content.

00:03:52: Exactly.

00:03:53: To rank in this new B-to-A-T-B world You have to structure your content from machines.

00:04:03: But, okay here's my pushback on that.

00:04:06: That creates a huge paradox.

00:04:09: if we strip all the pros away and just feed the chemical analyzer raw structured data how do we establish any actual brand authority?

00:04:18: Like why would an LLM choose our structure data over a competitor's structured data.

00:04:24: And that right there is where the human element becomes The Ultimate Tiebreaker.

00:04:27: Oh

00:04:27: really?

00:04:28: Yeah,

00:04:28: Justine Brownbridge shared some super compelling research on this.

00:04:32: She showed that LinkedIn Is actually currently the number one cited domain for professional queries across all of major AI platforms.

00:04:38: Wait!

00:04:39: LinkedIn is Number One Not Wikipedia or accompanying sites

00:04:42: Because AI models are inherently prone to hallucination.

00:04:46: So their developers tune them To actively seek out verified first-person human experiences to ground their answer.

00:04:52: Oh, that

00:04:53: makes total sense!

00:04:54: Yeah

00:04:54: so the winning formula isn't just technical schema.

00:04:57: it's embedding deep undeniable human expertise like real stories...real data inside that machine readable structure

00:05:05: which sounds great but trying to navigate that balance right now is causing so much friction in the industry.

00:05:10: Tom Darity made a really funny well funny but painful observation about where we are in the maturity cycle Right Now.

00:05:18: over The Acronyms Yes

00:05:20: He noted that we currently have twenty-four competing acronyms.

00:05:23: just for optimizing these engines, like agencies are pitching AEO, GEO, LMO and GIO.

00:05:30: It's ridiculous!

00:05:31: He argued.

00:05:32: the industry desperately needs an IE style standards body to step in and define a shared vocabulary

00:05:39: Because right now marketing directors are budgeting one acronym while their vendors sell them another one entirely

00:05:45: Exactly at The Mess.

00:05:47: Chaos around terminology usually just means a market is in transition.

00:05:50: And if we accept that AI search as fundamentally changing how buyers find your brand, where to look at the inverse too?

00:05:56: AI's also breaking how you're brand finds buyers right

00:05:59: outbound side

00:06:00: exactly which transitions us To The second major theme.

00:06:03: We saw the evolution of Outbound AI sales development reps or AISDRs.

00:06:08: I really want to dig into AISDRs heavily because this is the area where the marketing hype is just clashing so hard with actual revenue reality right now.

00:06:17: We are seeing teams fall into what's being called The Volume Trap.

00:06:20: The volume trap, let's break that down?

00:06:22: Yeah!

00:06:22: So Elisabeth Kuzeska and Alex Vaca shared this data set.

00:06:26: that illustrates it perfectly.

00:06:27: they looked at pure AIS DR.

00:06:30: fully autonomous outbound agents just set loose on an addressable market.

00:06:35: Just

00:06:35: blasting emails,

00:06:36: right?

00:06:37: And in one cohort these pure AI SDRs successfully booked a really impressive eight hundred and forty seven meetings in ninety days

00:06:46: which sounds amazing On paper it

00:06:48: does.

00:06:48: until you realize they only converted to pipeline at eleven percent.

00:06:51: ouch

00:06:52: Yeah contrast that with the alternative approach.

00:06:54: They tracked The human AI hybrid teams book significantly fewer meetings I think it was just three hundred and twelve, but their conversion to pipeline with thirty eight percent.

00:07:03: Wow!

00:07:04: Yeah so the hybrid team actually generated two point three times more real revenue while taking sixty-three percent fewer meetings.

00:07:12: Wait let's play out math on that pure AI approach for a second.

00:07:16: Eleven percent conversion on eight hundred and forty-seven meetings means that the human account executives were forced to sit through over seven hundred completely garbage meeting.

00:07:26: Exactly!

00:07:26: Like, The AI didn't create a robust pipeline.

00:07:29: it essentially executed a denial of service attack on its own sales team.

00:07:33: It burned AE time And probably burn bridges with a ton of unqualified prospects.

00:07:39: That is the hitter cost for volume trap.

00:07:42: The AI can execute the task.

00:07:44: It can send emails, but it completely lacks the capability to evaluate quality of outcome.

00:07:50: Trasoni actually shared a case study that reveals mechanics why hypermodel wins.

00:07:55: His team analyzed an AISDR system technically speaking flawless and found buying signals that generated grammatically perfect opening lines.

00:08:04: But uh...it was stuck at a dismal one point three percent reply rate.

00:08:09: One point three?

00:08:10: That's

00:08:10: terrible!

00:08:11: And the failure point was contextual judgment.

00:08:14: The AI doesn't inherently understand that a message being technically correct and a message been contextually appropriate to send today are two totally different things.

00:08:23: Yeah, because the AI does not intuitively know if the prospects company just laid off twenty percent of its staff yesterday or their industry is in the middle specific regulatory crisis.

00:08:33: Precisely It lacks the room reading skill.

00:08:36: So Sony's team implemented this really minor architectural change.

00:08:41: They added a human what they call the GPM system operator to the workflow.

00:08:46: Okay, and What did the operator do?

00:08:47: Their only job was to spend twenty seconds reviewing The context brief that the AI generated before authorizing the send.

00:08:55: That's it.

00:08:56: And by introducing that single point of human friction the reply rate rocketed from one point three percent To eleven point four percent.

00:09:03: just

00:09:03: for my twenty second review.

00:09:04: Just from them didn't change the underlying LLM.

00:09:07: They just placed human judgment right at the critical vulnerability points.

00:09:10: So let's contextualize that for the BWB marketers listening.

00:09:13: What does that twenty second human quality gate actually look like in practice?

00:09:18: Basically, The AI finds the lead it drafts the outreach based on intent data and then it queues up into a dedicated Slack channel.

00:09:26: Yeah exactly.

00:09:27: And then the Human Operator clicks the slack notification checks if tone is tone deaf to current events hits approve and moves On!

00:09:35: The human isn't drafting anything.

00:09:36: they are purely acting as an editor.

00:09:38: And that distinction between drafting and editing is really the core of The New Operating Model.

00:09:44: Arvi Karkhanji shared this highly ironic anecdote, which underscores it perfectly….

00:09:48: Oh I saw

00:09:48: one!

00:09:49: Yeah he overheard a conversation where an investor who was actively funding A very famous AI-SDR company was quietly trying to hire human SDRs for his own internal revenue team.

00:10:00: That is amazing.

00:10:02: Right, the investor driving the whole AI will replace sales narrative knew that to protect his own pipeline he needed personnel who could actually read the room pivot during live calls and navigate unscripted human objections

00:10:15: which brings a lot of clarity.

00:10:19: He compared the arrival of AISDRs to the ATM moment for the banking industry.

00:10:24: Oh, that's a great comparison.

00:10:25: It

00:10:25: really is like when ATMs rolled out The panic was that bank tellers would be entirely obsolete.

00:10:31: but ATMs didn't replace tellers.

00:10:33: they fundamentally altered the job description.

00:10:36: They handled a low-value manual task of counting twenty dollar bills, which freed the tellers up to focus on relationship banking loans and complex problem

00:10:46: solving.".

00:10:46: And Gilbert Kraliger noted the exact same dynamic in sales.

00:10:50: AI isn't replacing the strategic SDR.

00:10:52: It's just replacing the ninety percent of manual busy work they shouldn't have been doing in first place anyway like formatting CSV files, updating CRM fields writing generic follow-up templates.

00:11:03: those are tasks for machines.

00:11:04: strategy is for humans.

00:11:06: Okay.

00:11:06: so if adding a human operator fixes the conversion rates why are so many teams still failing to get ROI from these tools?

00:11:14: Which I think naturally transitions us into our third theme

00:11:17: The tech stack problem

00:11:18: Exactly.

00:11:19: The realization that you cannot build a smart hybrid SDR if your underlying tech stack is a fragmented mess, we are moving away from isolated tools and into orchestrated systems

00:11:31: because the tool fatigue is very real right now.

00:11:34: We saw Daria Novak map out an AI marketing stack recently That utilized twenty different disconnected tools.

00:11:41: Twenty?

00:11:42: Yeah, and Louis Cho showcased a twelve-tool stack just to run the solar marketing department.

00:11:47: I mean it looks super impressive on its slide but Sandeep Biladi provided a really necessary reality check on this.

00:11:53: What did he say?

00:11:54: He said collection of disparate apps is not a system.

00:11:56: Apps are waiting for their prompt A system connects data and actually compounds value.

00:12:01: Gladi pointed out that an AI without access to your proprietary data is just guessing based on internet averages.

00:12:07: Right So the critical infrastructure shift happening right now, Is the adoption of MCP or model context protocol?

00:12:12: We should

00:12:12: definitely explain MCP because it's literally dividing line between Amateur AI use and Enterprise AI operations in twenty-twenty six.

00:12:20: Go for it!

00:12:21: Think a standard large language model as incredibly intelligent intern.

00:12:25: They've read every book in the world, but they know absolutely nothing about your specific company.

00:12:30: MCP is the secure standardized pipeline that feeds that interned-your live CRM data... ...your historical campaign performance and customer analytics directly into their context window right before they answer a

00:12:43: query.".

00:12:43: It's the ultimate context injection!

00:12:45: Exactly so.

00:12:46: without MCP you're AI rights generic copy.

00:12:49: with MCP Your AI can pull last quarter's customer acquisition costs from your database and suggest budget reallocations based on historical win rates.

00:12:58: And that architectural shift dictates a massive change in marketing leadership?

00:13:03: Carolyn Healy argued, the CMOs who actually survive this transition must shift their core identity.

00:13:08: They have to move away from being campaign executors and become designers of agentic systems.

00:13:12: Designers of agentics system.

00:13:14: Yeah

00:13:14: You are no longer managing creative output you're managing data flows compliance guardrails escalation paths.

00:13:23: Okay, I look at the reality of most BtoB marketing departments and i have to ask.

00:13:29: Aren't we just risking the automation of our own bad habits here?

00:13:34: If a company's data hygiene is terrible Execute that broken process at light speed, isn't it?

00:13:47: Oh absolutely.

00:13:48: And that is the exact vulnerability Christine Royston and Lauren Layton both independently highlighted.

00:13:53: AI does not possess the ability to fix a broken workflow.

00:13:57: It merely exposes your operational gaps instantly.

00:14:00: It's a magnifying glass.

00:14:01: Exactly, Richard Jonkov brought some really sobering data to this discussion noting that currently only twenty-four percent of organizations actually have personalization operating at scale

00:14:10: Only twenty four percent?

00:14:11: Yeah most marketing teams cannot even activate the behavioral data they already possess.

00:14:15: So if you don't fix The Data Foundation and Decision Logic first AI will just amplify your fragmentation.

00:14:21: And managing That foundation requires serious dedicated human bandwidth.

00:14:27: Like, if you're a marketing director listening to this and your CEO is pressuring you just buy an AI agent to save headcount... You need to push back with the data Rutger Katz shared.

00:14:37: The Sastra data?

00:14:38: Yes!

00:14:39: He noted that Sastras runs twenty AI agents across their operations but the hidden reality is each one of those agents requires three-to four hours per week of senior operator time.

00:14:52: Twenty agents times three to four hours.

00:14:55: That is sixty-to eighty hours a week of senior strategic management just to keep the agents aligned with business reality.

00:15:01: It's practically two full time senior roles, right?

00:15:04: it isn't a task you could Just hand off to a junior coordinator.

00:15:07: Yeah pointed out that ninety percent Of AI implementations fail because leadership treats them as set and forget software.

00:15:14: an agent can give You a hundred and sixty eight hours of output A week but it demands rigorous human oversight To ensure that output Isn't hallucinated or off brand

00:15:21: which naturally brings us to our final theme.

00:15:24: If the system is architected correctly and data is clean, how do we fund and direct it?

00:15:32: The conversation has to move to strategy economics and culture required.

00:15:51: Adding one more user to the platform costs practically nothing.

00:15:55: That is why proceed.

00:15:55: pricing works, but AI economics operate on a completely different physical reality

00:16:01: because it requires raw compute power.

00:16:03: exactly every single prompt Every generation in every loop an agent runs burns API tokens which cost real money.

00:16:10: It's

00:16:10: basically the difference between renting server space to store static file and paying for a supercomputer To actively process a complex math equation every single time a user logs in.

00:16:21: That's exactly the dynamic.

00:16:23: If you try to run a twenty-twenty six AI motion using a twenty, twenty three size pricing model You will mathematically grow yourself into a loss.

00:16:32: heavy users won't just decrease your margins They'll actually bankrupt you.

00:16:36: Wow So go-to market teams are being forced To completely rethink their pricing structures.

00:16:41: and where they're competitive motes Actually lie The moat is no longer software features.

00:16:47: It is the proprietary data that your system compounds over time to make the compute more efficient

00:16:52: and That shift in where value is generated.

00:16:54: It applies directly to creative strategy as well.

00:16:57: Ryan law wrote a fairly painful truth for content marketers recently.

00:17:00: He basically said AI-generated content is now largely indistinguishable from human writing.

00:17:05: Yeah, that's a tough pill to swallow for a lot of people.

00:17:07: it

00:17:07: really is The quality gap that protected human writers a year ago has just gone.

00:17:11: consequently the value of pure production like Just generating grammatically correct words on a page as plummeted to zero.

00:17:18: Which is where Christenship for Barrett and Selkanca expanded the thought, they pointed out that speed without strategy just produces generic noise because AI amplifies the underlying foundation.

00:17:29: The floor-for-strategy has been raised significantly.

00:17:33: if your target persona Is vague?

00:17:35: And you're brand's point of view was dull the AI will simply allow You To produce That dull Generic Noise at an unprecedented Scale.

00:17:43: Strategy is literally the only differentiator left.

00:17:46: And implementing that strategy requires a totally different approach to talent.

00:17:50: Nicole Leffer shared this really powerful contrast regarding how leadership should view this transition, she described two hypothetical companies

00:17:58: Company A and company B.

00:17:59: Yeah so Company A used their new AI efficiencies to execute a classic cost-cutting maneuver eliminating one point five million dollars in marketing headcount.

00:18:09: of course the board was thrilled with immediate savings

00:18:11: But Company B took opposite approach right?

00:18:13: Exactly Company B retained their entire team and used the exact same AI expenditure to upskill their people.

00:18:21: By automating the busy work, they gave their existing humans the bandwidth to focus on deep market strategy in high leverage creativity.

00:18:28: And what was a result?

00:18:29: Company B ended up generating an extra ten million dollars in compounding annual recurring revenue.

00:18:34: see

00:18:35: that is the point right there.

00:18:36: The strategic objective of this era isn't cutting costs to save a million dollars.

00:18:42: it is compounding capabilities to generate ten million.

00:18:46: But, To become company B you really have address the human anxiety in this room.

00:18:50: Bjorn Raddy and Liza Adams both stress that AI adoption is not a technology challenge.

00:18:56: it's fundamentally a culture challenge.

00:18:58: Oh for sure!

00:18:58: You cannot mandate transformation.

00:19:00: Exactly It requires psychological safety.

00:19:03: If your team is terrified.

00:19:04: tool you are training them on is going to replace their salary next quarter, they will actively resist it.

00:19:09: Or even worse though use these miraculous paradigm shifting tools To execute the exact same legacy processes They were doing yesterday just slightly faster.

00:19:19: right leadership has to create an environment where It is genuinely safe to fail fast.

00:19:24: You need cross functional hackathons peer-to-peer learning and explicit guarantees that The goal is elevation not elimination?

00:19:34: All this complex machinery is just a lever.

00:19:36: The human, it's still the one deciding where to place the fulcrum.

00:19:40: That...is..a great way of putting it

00:19:41: Which actually leads us with final thought to mull over.

00:19:44: as we wrap up today We started by exploring Gillian Hart's concept Of B-to A-to B Business To AI To business.

00:19:52: Yeah Well if funnel is compressing And AI Is becoming gatekeeper for both sides Imagine the near future, where your highly optimized outbound marketing algorithm is simply negotiating directly with your prospects heavily guarded in bound procurement algorithms.

00:20:09: Oh wow!

00:20:10: Right if we reach a point of pure machine to machine transactions what happens?

00:20:14: To the role of human brand memory like how do you make an algorithm trust you?

00:20:18: and more importantly How'd he make it algorithm actually feel something about your brand?

00:20:22: If you enjoyed this episode, new episodes drop every two weeks.

00:20:25: Also check out our other editions on account-based marketing field marketing channel marketing martech go to market and social selling.

00:20:32: Thank You so much for joining us On This Deep Dive.

00:20:34: Don't forget to subscribe And we'll catch ya next time.