Best of LinkedIn: AI in B2B Marketing CW 25/ 26

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.

00:00:07: In calendar weeks twenty-five and twenty six.

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

00:00:18: You can find more info in description.

00:00:21: Welcome To The Deep Dive!

00:00:22: Thanks for having me always good be here.

00:00:24: Yeah

00:00:24: Of course Our mission today really decode the absolute top AI trends and the major shifts we've seen just sweeping across the B to D marketing space on LinkedIn over the past two weeks.

00:00:38: Right,

00:00:38: there's been a lot of chatter.

00:00:40: Oh!

00:00:40: A ton.

00:00:41: We're looking at The Good, The Bad And honestly very messy reality.

00:00:45: what happens when enterprise teams try scale this technology?

00:00:49: Yeah...and kick things off.

00:00:50: I want frame the core problem that you are seeing all these discussions.

00:00:55: Imagine you've just built this state of the art multi-million dollar water pipe, it's massive.

00:01:00: its incredibly efficient to can move.

00:01:02: I don't know.

00:01:02: a million gallons in minute

00:01:04: Sounds like total dream scenario for any operations or revenue team.

00:01:08: Like finally have that scale

00:01:09: Exactly.

00:01:10: You turn on valve The pressure builds and system just hums flawlessly Until walk till end.

00:01:18: Look down and realize you've just pumped a million gallons of toxic mud into a pristine swimming pool.

00:01:24: Oh, wow Yeah

00:01:25: right the infrastructure worked perfectly The volume was completely unprecedented but the actual output was entirely toxic.

00:01:33: That metaphor is exactly the crisis playing out right now with go-to market strategies, I mean teams are building these incredibly powerful AI engines but they're suddenly realizing that well automating a bad process just means you fail at the speed of light

00:01:45: precisely just feeling faster.

00:01:47: yeah so let's start right on the front lines without bound pipeline generation.

00:01:53: We saw this breakdown by Zach Daris showing how you can build a fully autonomous AISDR, sales development rep agent handling all cold outreach powered by Claude in under forty five minutes.

00:02:04: Forty-five minutes?

00:02:05: I know!

00:02:05: Forty Five Minutes to essentially build an employee.

00:02:08: and Yankos at Gigerich shared that his AIS DR booked one hundred and forty calls across the team in thirty days it had a thirty two percent reply rate without a single manual message.

00:02:21: That is just wild volume!

00:02:22: It really is, and then you have massive pushes like Taylor Herron running.

00:02:26: get this A million emails a month using intent data seeing click-through rates as high as thirty percent

00:02:32: right?

00:02:32: And when you hear numbers Like that the immediate reaction from literally any executive Is we need this deployed yesterday or were going out of business?

00:02:39: Oh for

00:02:39: sure.

00:02:40: total FOMO

00:02:40: Exactly.

00:02:42: But there's a hidden crisis here.

00:02:43: Alina Carnock shared a statistic that should honestly make every revenue leader hit the brakes, she noted that thirty three point four percent of teams at tried these AISDRs have already stopped using them.

00:02:55: hold on wait stop right they're.

00:02:56: if it tool books one hundred and forty meetings in a month without Any VP of sales is going to pop champagne.

00:03:04: You'd

00:03:04: think so, yeah Why

00:03:05: on earth would they turn a pipeline generator like that off?

00:03:08: That just doesn't make logical sense To me

00:03:10: because it creates this illusion of success.

00:03:12: you know Volume flatters the dashboard while entirely starving The pipeline of actual quality.

00:03:18: Oh...the

00:03:19: toxic mud from the water

00:03:20: pipe Exactly!

00:03:22: You can celebrate A massive spike in booked meetings but if those Meetings are with unqualified prospects who have absolutely zero intent to buy You haven't generated revenue, you've just wasted your sales team's entire quarter taking bad calls.

00:03:35: That makes a lot of sense.

00:03:36: yeah

00:03:36: Richard Dume has pointed out that while seventy-nine percent of companies bought AI sales tools recently only five percent say they actually work.

00:03:43: Five percent?

00:03:44: I mean...that is an abysmal success rate for enterprise software.

00:03:48: So what are the other ninety-five percent doing wrong?

00:03:50: Are we just like automating the worst parts of sales?

00:03:53: That exactly was happening.

00:03:55: Dume noted.

00:03:55: instead using AI to make outreach genuinely helpful or you know, highly researched teams just automated the nagging.

00:04:03: The nagging?

00:04:04: Right they trained supercomputers and say hey saw you download this white paper got fifteen minutes.

00:04:08: it's just faster spam!

00:04:10: Faster spam.

00:04:11: that actually reminds me of the story Eddie Reynolds shared.

00:04:13: It is hilarious but really kind of tragic for the state-of-the industry right

00:04:16: now.

00:04:17: Oh the cold call story.

00:04:18: Yes he got a cold call from a human SDR who was pitching an AI account research tool And the entire pitch was about how their software does deep research, so you don't have to send canned generic

00:04:31: templates.

00:04:32: The irony is already building.

00:04:33: Oh

00:04:33: it gets better!

00:04:43: So

00:04:46: let me ask you this.

00:04:47: Are we just strapping a jet engine to a broken bicycle here?

00:04:50: Like if the underlying message and strategy are fundamentally bad, aren't we just failing faster?

00:04:56: that's The perfect analogy.

00:04:57: I mean A smarter language model doesn't fix a broken outbound strategy.

00:05:02: timing does right

00:05:03: timing And this is where we sort of transition from what teams are doing wrong To how to actually fix it.

00:05:09: a revenue leader going by SKB shared a really brilliant insight on this.

00:05:14: They completely stopped optimizing their email copy and started optimizing purely for intent.

00:05:20: What does that look like in practice?

00:05:22: Well, they pointed there AI model at three hundred specific buying signals things like executive job changes fresh funding rounds or prospects engaging with competitors online.

00:05:33: Oh I see.

00:05:34: so instead of telling the AI rate a really clever e-mail to this random list Only look at these three hundred companies and alert me the second they show a specific behavior.

00:05:45: Exactly, by letting the AI focus strictly on intent in timing They booked eight meetings in fourteen days.

00:05:51: The message wasn't poetry didn't need to be.

00:05:53: it just arrived with the exact second that buyer actually needed this solution.

00:05:56: That makes so much sense.

00:05:57: It really separates the robotic busy work from the actual human relationship.

00:06:01: Daniel Lutio Katzen & Rob Cook both argued this beautifully In their posts.

00:06:07: AI should own the things humans naturally hate and are just generally bad at, right?

00:06:12: Mass prospecting deep data research CRM updates summarizing transcripts.

00:06:17: The administrative garbage basically

00:06:19: garbage.

00:06:20: but humans must own the parts that actually move complex deals-the emotional discovery ,the nuance negotiation building actual trust.

00:06:29: Right you don't use AI to replace the salesperson's judgment.

00:06:33: You use it to clear the desk so they can actually apply that judgement.

00:06:36: Okay, so if outbound is how we push-out to buyers and as we just established it's becoming this massive tidal wave of automated noise.

00:06:43: buyers are naturally going put up walls right?

00:06:44: On

00:06:44: buyer.

00:06:44: solutely they're already doing that.

00:06:46: Yeah!

00:06:46: They're gonna rely more on pulling us in their own terms.

00:06:49: But this is where the dynamic completely shifts, because traditional SEO search engine optimization is fracturing.

00:06:56: We're moving headfirst into the era of AI Search or AEO Answer Engine Optimization

00:07:01: and This Is a huge seismic shift in how discovery works.

00:07:05: Owolani framed the mechanics of this perfectly.

00:07:08: for what twenty years?

00:07:11: Traditional Google SEO asked one structural question Who ranks number one?

00:07:16: Right, based on page architecture and hyperlinks.

00:07:18: Exactly.

00:07:19: but AI search engines you know like JADGPT or Proplexity they don't care about page rankings And AI evaluates semantic relationships.

00:07:28: to ask a completely different question who can I trust enough to recommend?

00:07:31: And the scary part is that you can be completely winning the old game and entirely losing the new one without even realizing it.

00:07:38: Yes, exactly.

00:07:38: Like Connor Gillivan he ran his company Ecom Balance through a new AEO assessment... ...and he scored two out of five!

00:07:45: Ouch!

00:07:46: Yeah, what he found was fascinating.

00:07:48: when buyers asked general informational questions like just browsing for concepts His company showed up in the AI answers But the second a buyer asks an AI decision stage question, like say comparing ROI between vendors.

00:07:59: Ecom balance completely vanished.

00:08:01: Wow Yeah they were visible when buyers are browsing and entirely invisible When They Were Actually Ready To Buy.

00:08:06: That is A brutal blind spot to have as a marketer.

00:08:10: Matt Shealy laid out The New Scorecard For This.

00:08:12: You Can't Just Check Your Keyword Rankings Anymore.

00:08:15: you Have To Ask A Whole New Set Of Questions Like Wipe Like, are we named in the AI's answer?

00:08:20: Yeah.

00:08:20: Which competitor gets named instead of us?

00:08:23: which source data caused The AI engine to trust them?

00:08:26: and most importantly can We influence that underlying Source?

00:08:29: I want to push on That for a second.

00:08:31: it sounds like optimizing for, well a digital reputation system now.

00:08:35: You can't just hack your way to the top of Google with five thousand word listicles and keyword stuffing anymore?

00:08:40: Not at all!

00:08:41: you actually have to be recognized as The Authority by the AI's training data

00:08:45: Exactly And we have the data to prove how this mechanism Actually works.

00:08:50: Kyle Atwater Morley analyzed what genuinely drives visibility in large language models.

00:08:56: For years marketers obsessed over backlinks, you know other websites linking to yours.

00:09:00: Right

00:09:00: the golden goose of SEO.

00:09:02: But in AI search back links are a terrible predictor of visibility.

00:09:06: The correlation is under point three euro.

00:09:08: Do You Know What Actually Forces an AI To Recommend You?

00:09:11: Brand Mentioned In YouTube Impressions.

00:09:13: They Have A Correlation Between Point Five Zero And Point Seven Four.

00:09:17: Oh Wow Because the AI isn't just crawling a web of links, it's reading the internet actual conversations.

00:09:23: It is looking at entity association

00:09:25: Exactly

00:09:25: And Alex Alene confirmed this.

00:09:27: He noted that LinkedIn actually second most cited domain in all of AI search Acquiring an eleven percent Of all answers.

00:09:34: But I mean Just getting your name dropped Isn't enough right?

00:09:37: I saw Andrew Yon's data on Helfords, the massive UK auto parts retailer.

00:09:42: This is a great example!

00:09:43: Yeah they are mentioned in over eighty eight percent of AI answers about their category and the AI clearly knows who they are.

00:09:51: but Helford actual website it only cited as the source link.

00:09:54: three point five percent at that time.

00:09:56: yeah thats what we call the citation gap.

00:09:57: The AI understands semantically that Helfards an authority so mention them into text.

00:10:02: But Halford's own content isn't structured to answer the user-specific prompt, so the AI has link like a third party reviewer or forum.

00:10:11: actually prove its point.

00:10:13: Gaitano Nino Dinardi pointed out that most of these AEO problems aren't technical SEO problems at all.

00:10:19: They are fundamentally positioning and category fit

00:10:22: problems.".

00:10:22: Yes,

00:10:23: and Zach Peruch tied this back to a fundamental marketing concept from that twenty-ten book How Brands Grow?

00:10:30: He pointed out.

00:10:31: the traditional search is driven by keywords.

00:10:33: right of buyer types increase web traffic.

00:10:36: Sure, but AI search is driven by category entry points.

00:10:39: buyers literally type their entire messy complex situation into the prompt box They'll type.

00:10:45: I can't tell if my recent traffic drop Is due to a Google algorithm change?

00:10:49: The rise of ai overviews or If my content is just bad.

00:10:53: how do i audit this

00:10:54: right conversational

00:10:56: Exactly.

00:10:56: And if your content isn't anchored to those highly specific real-world buyer situations, the AI will completely ignore you no matter how many times we use keyword traffic.

00:11:05: Which is why Garrett Sussman has warned that this is a zero sum game.

00:11:08: You can just have one junior SEO manager running random prompt checks once a month.

00:11:13: Definitely not.

00:11:14: When brand gets included in an AI answer another brand usually get's dropped.

00:11:18: Sussman noted that Reddit citations jumped four point four times after a single chat GPT update.

00:11:24: Wow, four-point-four times!

00:11:26: Yeah the underlying algorithms are wildly volatile.

00:11:30: you need a dedicated cross functional team constantly monitoring this.

00:11:34: Okay, let's connect these dots.

00:11:35: To execute this hyper-targeted outbound we talked about earlier and to manage this massive new AI search visibility.

00:11:42: companies are deploying AI agents autonomous software that can execute tasks

00:11:46: right.

00:11:47: but here is the mass of roadblock.

00:11:49: if you take an incredibly smart agent And drop it into a fragmented messy tech stack You don't get efficiency.

00:11:55: You just get automated chaos.

00:11:57: welcome to The Messy Middle.

00:11:59: This is where the theoretical dream of AI violently collides with a reality of enterprise data.

00:12:04: Yeah,

00:12:04: Lisa Sharapata calls it random acts of AI.

00:12:07: instead of a rogue human sales rep sending a bad email You now have a rogue AI agent writing a pitch at eleven p.m.. Updating a slack channel but never logging the interaction in the CRM.

00:12:17: It's automation theater.

00:12:18: exactly The work is happening But the business has no record of it.

00:12:22: Henry Shuck diagnosed exactly why this architecture is feeling Enterprise.

00:12:27: AI has essentially conquered software engineering and customer support.

00:12:31: Why?

00:12:32: Because that data is highly centralized in clean systems like GitHub or Zendesk, but AI hasn't cracked go-to market yet.

00:12:40: because GTM data You know, every seals call.

00:12:43: Every marketing email and website visit it's entirely scattered across twenty different platforms right?

00:12:48: He gave a brilliant example of how this breaks down mechanically.

00:12:51: an AI agent gets completely paralyzed if your CRM lists PR tourism as a cold prospect but you're billing software lists Quiderico tourism agency as an existing paying enterprise customer.

00:13:04: Because it's machine, It lacks common sense if the AI doesn't explicitly know they're the exact same entity and might just blast a twenty percent discount cold email to an Enterprise account that is already paying you full price.

00:13:14: exactly

00:13:15: Majovo he broke this architecture problem down into four distinct layers layer one the large language model, The Brain of the operation.

00:13:23: Layer four is the execution tools like Salesforce or HubSpot, The

00:13:26: Hands.".

00:13:27: Okay what about middle?

00:13:28: Well the middle layers are completely missing in most companies.

00:13:31: we lack layer two which is shared context a unified definition to your ideal customer profile and account history.

00:13:38: And We Lack Layer Three Which Is Orchestration...the actual logic deciding which system fires when

00:13:44: okay.

00:13:44: so To use an analogy, letting an AI loose in a fragmented CRM is basically like putting a brilliant Michelin star chef in a commercial kitchen.

00:13:54: But the ingredients are hidden inside the walls The labels aren't a foreign language and half of the produce has expired.

00:13:59: That's exactly what it's like!

00:14:01: The

00:14:01: Chef Is A Genius but the meal will still be garbage!

00:14:04: Hundred percent...the reasoning engine just starved with context.

00:14:08: Benjamin Reed argued that to fix this, AI needs a unified person graph.

00:14:13: This means creating one single timeline and a complete unified customer record that syncs across every single system in real time.

00:14:20: Right, otherwise marketing sees one version of the customer sales sees another And the

00:14:24: AI is just making highly confident guesses.

00:14:27: And guessing at scale Is incredibly dangerous.

00:14:31: We saw Guillaume Baleel bring up A sobering warning from The Bank Of England regarding Autonomous Agents In The Financial Sector.

00:14:37: Oh That Was Fascinating.

00:14:38: Yeah They Noted That Simply Having A Human In The Loop Which is, let's be honest everyone's favorite safety buzzword right now isn't enough.

00:14:46: If an agent operates inside critical workflows you need hard constraints circuit breakers rollback plans and actual audit trails.

00:14:55: Because think about the speed we are dealing with.

00:14:58: if an agent hallucinates and overwrites a thousand CRM records in three seconds A human in the loop can click undo fast enough.

00:15:05: No way.

00:15:06: Human oversight is completely useless if the human doesn't have time for architectural tools to reverse machine's action.

00:15:12: Amos Bar-Joseph offered a massive paradigm shift here, he said we need stop thinking of AI as an AI CMO or an AI sales rep because frankly AIs are terrible marketers.

00:15:22: it does not understand human nuance at all.

00:15:25: what AI actually is is brilliant software engineer.

00:15:28: Yes you don't buy off the shelf digital worker and tell them just do marketing You use the AI write the custom code and build the exact software plumbing.

00:15:39: you need to bend your tech stack around your specific unique go-to market strategy.

00:15:45: You

00:15:45: call the shots, you decide what good looks like... The AI just builds the infrastructure to make it happen instantly which brings us to the final I think most important piece of this entire puzzle.

00:15:55: If agents are doing data plumbing if they're sending hyper targeted emails or synthesizing search answers What exactly is the role of The Human Marketer?

00:16:06: It's a million dollar question.

00:16:07: Right, I have to ask.

00:16:08: as content generation becomes infinite and essentially free Doesn't human taste become the only real moat A brand has left?

00:16:16: it absolutely does.

00:16:17: Louise Brace had a brilliant reflection on this exact existential crisis.

00:16:21: She said we're starting to confuse information with expertise.

00:16:24: Oh that's good right.

00:16:25: just because you can type a prompt into an AI And get it to spit out at twenty page marketing strategy doesn't make you a marketer.

00:16:33: Yeah, anyone can retrieve information now.

00:16:36: Exactly the real defensible value comes from the judgment behind the prompt knowing what specific questions to ask understanding the cultural context The AI lacks and honestly having the expertise.

00:16:51: Drew Nyser was very passionate about this on LinkedIn.

00:16:54: He argued that AI assisted shouldn't be treated as a scarlet letter?

00:16:58: Yes, there is a mountain of AI-generated slot flooding the internet right now but let's be honest... There is also a Mountain Of Thoughtless human only content That reads like it was written by robots.

00:17:10: Oh

00:17:10: for sure

00:17:11: The tool isn't the villain!

00:17:12: The lazy thinker is.

00:17:14: discernment Is the differentiator

00:17:16: And that discernment is what drives actual measurable business growth.

00:17:21: Marcus Pritchard from Procter & Gamble shared a perfect real-world example of this human moat in action.

00:17:26: Oh, the dish soap story?

00:17:28: Yes!

00:17:28: They

00:17:29: had successfully launched Don Powerwash here in

00:17:31: U.S.,

00:17:31: and they wanted to adapt that success for their ferry brand.

00:17:35: Now an AI could easily analyze the demographic data and spit out a dozen clever taglines.

00:17:40: But it took human observation to realize a crucial cultural nuance.

00:17:46: British people have a deeply ingrained ritual of soaking their dirty dishes in the sink before washing them.

00:17:52: Right, An AI sitting at a server farm can't observe a cultural physical ritual In a kitchen in London.

00:17:58: Exactly That distinctly human insight led to skip-the-soak campaign which drove double-digit growth for the brand.

00:18:06: AI accelerated the execution of the campaign, but human empathy and observation found the core insight that actually made it

00:18:12: work.".

00:18:12: But there is a very real trap here isn't there?

00:18:14: I'm looking at Eric Malillo's post about a recent BCG study... It found when people use A.I.. For complex problems That they A.i.. Actually handles badly.

00:18:24: The A. i sounds super confident in its output.

00:18:27: The humans relying on it are nineteen percent less likely to spot mistake than if they hadn't used tool.

00:18:33: Yeah, we get lulled to sleep by the confidence of machine.

00:18:36: Which is why taste and discipline from failure are so critical.

00:18:40: Kevin McGrew summed up new operating system for marketers perfectly.

00:18:43: It's just four words Try Fail Improve Succeed.

00:18:48: I like that.

00:18:49: Most teams try fail And repeat failure at scale because they let AI run on autopilot.

00:18:56: AI can take this shot and move incredibly fast but only the human can look at their failure, study exactly what broke in process adapt strategy and communicate nuance to market.

00:19:07: To bring it all way back your opening metaphor we have to be the ones analyzing mud in swimming pool figuring out where pipe broke fixing water source.

00:19:16: we cant just stand there marveling how fast pump is running.

00:19:20: if connect these themes AI search discovery, broken tech stacks and human judgment.

00:19:25: And project this out a few years we arrive at really fascinating scenario.

00:19:29: What's that?

00:19:30: what happens when an enterprise buyer deploys an AI agent to evaluate vendors... ...and it meets your AI seller agent?

00:19:36: Oh wow!

00:19:37: Agent-to-agent.

00:19:38: Yeah

00:19:39: the buyers AI will parse your data perfectly.

00:19:42: Your sellers AI will pitch perfectly When they algorithms perfectly cancel each other out.

00:19:46: The only tiebreaker left Will be the offline Human Trust your brand has built in real world.

00:19:52: the digital reputation system eventually just points right back to your human reputation.

00:19:57: That is definitely something to really chew on as we navigate this transition.

00:20:00: Absolutely!

00:20:01: Well, if you enjoyed this episode new episodes drop every two weeks.

00:20:05: also check out our other editions on account based marketing field marketing channel marketing martech go-to market and social selling.

00:20:13: thanks for joining us on this deep dive into the future of BDB Marketing.

00:20:18: Thanks For Listening.

00:20:19: don't forget to subscribe And We'll See You Next Time.