Best of LinkedIn: MarTech Insights CW 24/ 25

00:00:00: This episode is provided by Thomas Allgeier and Frennus, based on the most relevant LinkedIn posts about MarTech from calendar weeks twenty-four to twenty five.

00:00:08: Frenness is a BtoB market research company that supports enterprise marketing teams in unlocking full potential of their customer data with help of AI.

00:00:16: You can find more info in description.

00:00:22: Don't miss out on BtoB Ignite, twenty-twenty six taking place one July.

00:00:26: Twenty twenty six in London.

00:00:27: the event brings together leading voices and b to be marketing for a full day of keynotes practical workshops pure networking hands-on problem solving sessions across demand brand leadership and tech.

00:00:38: find registration link description.

00:00:40: so I'm in June.

00:00:43: Someone reportedly offered one point two billion dollars just to know exactly who your customers are

00:00:48: which is Just a staggering amount of money for identity track

00:00:50: right.

00:00:51: I mean not to buy new software platform Not to get some slick, New marketing dashboard.

00:00:55: Just own that underlying connective glue.

00:00:57: You Know the stuff The tracks identity across the internet.

00:01:00: so welcome To the deep dive.

00:01:01: yeah glad to have you all here because today we Are basically tearing down the architecture Of b-to-b Marketing as you know it

00:01:08: Exactly Because the foundation Is well It is fundamentally cracking And it is being rebuilt in real time.

00:01:14: It really is and that one point two billion dollar number.

00:01:18: It is still ringing in my head, honestly We have spent the last few days just combing through the top martech trends that surfaced across the industry In calendar weeks twenty four and twenty five.

00:01:29: there's a lot to sift though definitely

00:01:32: But our mission for this deep dive is to cut through all at hype you know, to figure out the actual mechanics of what is happening beneath the surface.

00:01:40: Because

00:01:40: we are looking at a complete rewiring of customer data and AI right now

00:01:44: Exactly.

00:01:45: So let's just start with the tectonic shift that triggered all this which is uh... The Data Warehouse essentially swallowing marketing stack whole.

00:01:53: Yeah!

00:01:53: That was great way to put it.

00:01:54: I mean for very long time the Data Ware House was just passive storage.

00:01:59: It was the basement archive where your data went to sit-and-die

00:02:02: Right Just bunches SQL tables.

00:02:04: nobody in Marketing ever looked

00:02:05: Exactly, but mechanically that is changing fast.

00:02:08: We are seeing the warehouse become the active operational center of gravity.

00:02:13: Which

00:02:13: brings us to what Katie you and posted about right?

00:02:15: The day as twenty-twenty six summit.

00:02:18: Yeah So at that Summit Databricks unveiled customer Lake To an audience of like thirty thousand leaders.

00:02:24: That

00:02:24: isn't massive it

00:02:25: Is And they're billing It As An agentic CDP That Is natively Embedded Right Inside the Databrick's lake house.

00:02:32: Okay, I need to pause you there because the terminology gets really dense quickly.

00:02:36: When you say an agentic CDP embedded natively mechanically how is that different from what an enterprise marketing team actually does today?

00:02:44: Well today they buy a standalone customer data platform of CDP and they just pipe their data into it right.

00:02:49: They set up the connectors.

00:02:50: They've piped it over

00:02:52: but that piping Is The entire problem.

00:02:54: in the traditional model your data lives In a warehouse.

00:02:57: To use It Your cdp has to make API calls to copy that data, pull it over sync it and then store a duplicate version of it in its own silo.

00:03:06: And your email platform probably copies again right?

00:03:09: Exactly!

00:03:10: It is incredibly inefficient...it has prone-to latency....and you know..it's total nightmare for security.

00:03:15: Oh I bet.

00:03:17: So the shift with something like Customer Lake is what we call zero copy architecture.

00:03:21: You are no longer moving the data To the marketing tools.

00:03:25: You bring the tools to the data

00:03:26: precisely.

00:03:27: you are bringing the identity resolution The audience building directly, too where?

00:03:32: Data already lives.

00:03:34: so no copying No syncing delays.

00:03:36: So just use an analogy here at the old CDP model is like having a translator who has to constantly run back and forth between two different embassy buildings just to relay a message.

00:03:46: That

00:03:46: is exactly what it's like!

00:03:47: And the Lakehouse model simply puts all of the ambassadors in single room.

00:03:51: speaking a universal language, no one has to leave that building...

00:03:54: ...that captures mechanics perfectly yeah?

00:03:56: It really explains Seeger Dirick reaction into this whole launch-

00:03:59: What did he say about

00:04:00: him?!

00:04:00: He pointed out that CDP as software category was actually never product problem always data architecture problems in disguise.

00:04:09: Oh wow That is a sharp way to look at it.

00:04:11: Right, I mean marketing teams spent years and literally millions of dollars buying standalone CDPs To try and fix fragmented messy data.

00:04:22: Yeah bolting a shiny interface on top of a broken foundation

00:04:25: Exactly And by putting the cdp inside The lake house data books essentially arguing that you just cannot fix marketing without fixing the underlying data architecture first.

00:04:35: Like putting a supercomputer in a car with no wheels?

00:04:39: As Seegert said, you are just running autonomous campaigns confidently wrong faster.

00:04:44: Confidently Wrong exactly!

00:04:45: But

00:04:45: wait isn't that incredibly self-serving for Databricks to say like hey marketers To fix your campaigns You simply must buy our massive enterprise data infrastructure?

00:04:54: Well sure there's obviously a motive right?

00:04:56: Is it actually better for the marketer or is it just gravity winning out?

00:04:59: For the big data providers.

00:05:01: I know David Chan weighed in on

00:05:02: this.

00:05:02: he did and chan makes a really vital distinction here.

00:05:06: He noted that the CDP isn't a dying category waiting to be commoditized.

00:05:10: Okay, so it isn't going away.

00:05:12: no It has simply become a capability So deeply foundational That embedding into the data infrastructure is Actually proof of its ultimate importance.

00:05:22: I see It graduated from being a tool to be infrastructure.

00:05:25: Exactly!

00:05:26: However, to your point about Databricks just pushing its own narrative... David Rab offered very sharp contrary perspective.

00:05:33: Oh interesting what's his take?

00:05:34: He argues the data warehouse simply cannot be the sole source of truth for marketing.

00:05:39: Wait why not?

00:05:40: if this zero copy architecture is so efficient and all historical data are sitting right there Why wouldn't it be the source of Truth?

00:05:47: Because of latency in context.

00:05:50: Think about the mechanics of a real-time customer interaction for a second.

00:05:53: Okay,

00:05:53: like someone browsing your pricing page?

00:05:55: Exactly!

00:05:56: A buyer is on your website right now clicking around while simultaneously you're inventory levels for specific service tier are fluctuating.

00:06:03: Right and that split seconds contextual data doesn't live neatly in warehouse.

00:06:07: yet

00:06:08: exactly.

00:06:08: it takes time to get there.

00:06:10: Rob argues routing all those real-time unstructured signals through a central warehouse first, just to make a split second marketing decision.

00:06:18: Is like routing local traffic though national highway system?

00:06:21: Yes

00:06:22: exactly it is just too slow he believes.

00:06:24: the true heart of Martek isn't data store It's the orchestration layer

00:06:29: The system that can actually access data wherever it sits to make immediate decisions.

00:06:34: Exactly Look, I see the logic there.

00:06:36: if you rely purely on the warehouse You might have a perfectly unified historical profile But you completely missed the context of what the buyer was doing literally three seconds ago.

00:06:47: It is a blind spot.

00:06:48: So for the enterprise teams out there trying to build their roadmaps, I think Bindi S Raj had some really good advice.

00:06:54: She said you have to stop evaluating these tools in isolation.

00:06:57: Yeah You can no longer just run an RFP For standalone CDP.

00:07:00: Exactly!

00:07:00: You have to evaluate The full data-to-activation stack The warehouse The orchestration The channels All as one single architectural decision.

00:07:09: And that architectural Decision brings up A massive operational hurdle Frankly

00:07:13: Which Is?

00:07:14: Well

00:07:15: If you agree that the Lakehouse is The New Foundation and you aren't copying data out to external tools anymore, how do you actually identify who the person is?

00:07:27: Because data doesn't naturally know that user one two three browsing your site on a mobile device.

00:07:33: Is the exact same person as John Doe in your CRM

00:07:36: which brings us right back To That One Point Two Billion Dollar Number we started with exactly.

00:07:40: We have all this unified Data sitting In A Governed lake house But without a way to map the identities across different devices and touch points, I mean it is just very fast.

00:07:50: Very expensive filing cabinet.

00:07:51: A filing cabinet that can't read its own files?

00:07:54: Yeah And perfectly explains explosive news that Hightouch reportedly made bid for LiveRamp's ID assets.

00:08:01: Right.

00:08:01: Martin Keene called this move an active existential panic didn't he?

00:08:05: He did To understand why you really have.

00:08:07: look at the mechanics of identity resolution You see Hightuch known as reverse ETL company

00:08:13: meaning they help move data out of the warehouse.

00:08:15: Right, but if Databricks is building customer leg so Data never has to leave in first place The entire value proposition for moving data kind just dissolves.

00:08:24: So Hightouch naturally looks to acquire a real-time deterministic ID graph because They have own something foundational

00:08:32: Precisely!

00:08:33: Matthew Niederberger even noted that this specific scramble To own infrastructure really defined the month June.

00:08:40: Let's actually break down that term for a second.

00:08:43: Deterministic ID graph, because that is essentially the core of the one point.

00:08:47: two billion dollar valuation right?

00:08:49: Right

00:08:49: it is The Holy Grail

00:08:50: Because A probabilistic graph just sort of guesses who user Is based on IP addresses browser cookies behavioral patterns.

00:08:58: It isn't educated.

00:08:59: guess

00:08:59: Yeah and his fuzzy math.

00:09:00: But a deterministic graph uses hard identifiers like I hashed email address from me user Who Actually logged into a portal it knows.

00:09:08: For a fact to you are

00:09:09: Exactly.

00:09:10: And as third-party cookies vanish, deterministic matching is basically the only reliable way to connect the dots anymore.

00:09:16: So the whole scramble was about bringing that deterministic match directly into data?

00:09:21: Yep!

00:09:22: Marci Von Tadelli highlighted Axiom.

00:09:24: Real ID is now natively available inside Databricks Customer Lake.

00:09:27: Oh, natively...that's

00:09:30: huge!

00:09:30: Very huge and Dave Lambert noted that Iterable is a launch partner too.

00:09:35: The mechanical goal of all this is identity resolution without ever exposing PII.

00:09:40: Personally identifiable information,

00:09:43: right?

00:09:43: You don't have to send your customer list out two a dozen different external vendors to be matched anymore.

00:09:49: the matching algorithm literally comes to your secure database

00:09:53: which Is where things start to sound I dunno A little dangerous for legacy Martek platforms.

00:09:59: How so?

00:09:59: Well,

00:09:59: think about it.

00:10:00: if the data isn't moving and the identity matching is happening natively inside The Lake House what exactly is holding the marketing stack together?

00:10:07: Ah right...the

00:10:09: glue!

00:10:09: Yeah I'm thinking of Mike Berry's observation here.

00:10:12: he pointed out that historically marketing teams bought CDPs because they desperately needed a platform to hold everything together.

00:10:18: He

00:10:18: essentially paid six figures just by the connective glue

00:10:21: Exactly.

00:10:22: But if AI is fundamentally replacing that connective work, I mean historically you needed a CDP to map the data fields build the logical rules for an audience stitch the API connections to your email platforms.

00:10:35: but today in AI agent could just sit on top of a resolved customer store In The Lakehouse and do all of that

00:10:41: natively right.

00:10:42: it can interpret a natural language request like Find me All Enterprise buyers who looked at pricing yesterday And

00:10:48: It Just Builds The Audience Via SQL assembles the creative and pushes it out.

00:10:53: The

00:10:53: manual stitching that a CDP used to do is now just a task an AI can execute natively, It Is Wild!

00:11:01: So this

00:11:01: six-figure hidden cost of marketing was literally just the stitching?

00:11:05: Pretty much.

00:11:05: And if A.I does that stitching for fractions of a penny Now...the entire battleground shifts.

00:11:11: Christopher Marriott summed us up so well.

00:11:13: What did he say?

00:11:13: He said the strategic focus for organizations is totally moving.

00:11:17: We use ask who owns the customer profile?

00:11:19: Now, the real question is who can act on customer intelligence most effectively with minimal friction?

00:11:24: That distinction is huge.

00:11:26: The difference between merely storing intelligence and actively executing on it.

00:11:31: that Is the dividing line for the next era of BDB marketing.

00:11:34: We are moving from human marketers manually clicking buttons in building logic trees to autonomous agentic AI workflows

00:11:43: which means we're talking about Building the Customer Brain.

00:11:46: Yes.

00:11:47: I am honestly fascinated by this concept of agentic marketing, mostly because B-to-B feels weirdly behind the curve

00:11:53: here.".

00:11:53: It really does!

00:11:54: Arjun Pillai made a sharp observation about that – he pointed out that capital has completely flooded into AI for the sales side… Oh

00:12:01: absolutely!

00:12:02: Endless AI, sales development reps, outbound email bots, automated dialers... Right

00:12:07: but AI seems to be skipping the top of the funnel in B-to-B marketing entirely.

00:12:11: Brand interaction and complex buyer journeys have seen almost zero agentic innovation.

00:12:15: Why do you think that is?

00:12:17: Well, it really comes down to the mechanical difference between generating and reasoning.

00:12:21: John Miller broke this down recently.

00:12:23: Okay

00:12:23: tell me more.

00:12:24: So generative AI.

00:12:25: You know The large language models we're all used too.

00:12:27: It has made content personalization incredibly cheap.

00:12:30: You can use an LLM to rewrite a cold email intro for hundred different prospects and it costs practically nothing.

00:12:37: Right, but mechanically that LLm is just predicting the next most likely word in a sequence.

00:12:44: It's generating text...it isn't reasoning

00:12:47: Exactly!

00:12:47: Its essentially just massive mail merge on steroids.

00:12:50: Yeah its sounds personalized But there absolutely no strategic intelligence behind why this specific email was sent at such time.

00:12:59: precisely True one-to-one personalization requires an AI that can actually reason.

00:13:04: Miller argues that a reasoning AI must simultaneously solve for four variables, which are the offer of channel timing and content.

00:13:12: Okay wow!

00:13:13: And traditional rules birth systems like say a Marketo logic tree.

00:13:16: they could never weigh all those real world signals at once?

00:13:19: Never.

00:13:20: but an agentic AI.

00:13:21: look in buyers history recent website activity their company's funding news dynamically decide what to do.

00:13:29: This person shouldn't get the standard nurture email today.

00:13:32: They should get a personalized case study delivered via a LinkedIn message next Tuesday.

00:13:36: Exactly!

00:13:37: That requires complex decision-making, not just text generation.

00:13:42: I see the challenge there though because if you want an AI to make that level of strategic decision, The underlying data has to be absolutely flawless.

00:13:50: It has to perfect

00:13:51: which connects To a brilliant point Gerard Hilo made about where marketing budgets are just being completely wasted right now.

00:13:58: Oh yeah Where they wasting it

00:14:00: he noted?

00:14:00: That every marketing team wants to buy the shiny ai tool that writes the campaign copy But generating assets is abundant.

00:14:08: Now its practically free.

00:14:11: The actual bottleneck, the scarce resource you should really be funding is having clean resolved data to know if a campaign should even run in first place.

00:14:20: Spending your budget on the generation layer while completely ignoring the data foundation is just backward

00:14:24: Totally backwards.

00:14:25: If you bolt an incredibly fast reasoning AI agent onto fragmented broken data it is just gonna produce confident, expensive nonsense at an unprecedented scale.

00:14:37: Confident,

00:14:38: expensive non-sense?

00:14:40: It will confidently email the exact wrong offer to your absolute biggest client!

00:14:45: That's

00:14:45: a perfect way to describe bad automation and really highlights why Nabindupal's concept of customer brain is so crucial right now.

00:14:54: Yeah he drew hard mechanical line between personalization & decisioning didn't he?

00:14:58: He did...he said personalization just making an experience relevant.

00:15:02: You know, like putting their company name in the subject line...

00:15:05: Right?

00:15:05: The basics!

00:15:06: But a customer brain handles real-time decisioning long before personalization is even applied

00:15:11: Because mechanics of customers' brains involve continuously ingesting contextual signals and historical outcomes

00:15:17: right?

00:15:17: Exactly To determine next best action on strictly mathematical level.

00:15:22: If Personalization is the wrapping paper Customer Brain is the intelligence deciding what actually goes into box

00:15:28: And whether it should be sent at all and who the courier should be.

00:15:31: Exactly!

00:15:33: However, Nabandu also warned that enterprises are not fully ready to just build their own AI brains from scratch.

00:15:40: It requires intense data governance.

00:15:42: it is NOT about buying a new software license.

00:15:45: The operational reality is definitely daunting but I mean THE VISION for where this leads IS JUST WILD.

00:15:52: Kaurav Vasivada & Dave Burkhed painted picture of near future.

00:15:56: that forces us rethink the very definition.

00:15:59: Oh yeah, they argue that we need to stop thinking of MarTech as a passive dashboard.

00:16:03: Right for the last twenty years marketing meant a human logging into a CRM, logging in on social tools and manually pulling levers.

00:16:11: Burkhead says the era of the marketer login is ending

00:16:15: So he needs to treat martech an active digital workforce instead

00:16:18: Mechanically.

00:16:19: what burkhead is describing it's intelligent layer which sits above traditional stack The tools we use to log into every day the email senders, the analytics dashboards.

00:16:28: They're basically becoming invisible plumbing.

00:16:30: And then new primary interface is the AI agent itself.

00:16:34: Right you interact with the agent layer giving a strategic objective and the agent goes down Into the infrastructure queries the data builds the audience and executes all the tasks across the different platforms

00:16:45: Which is crazy.

00:16:46: But we are actually already seeing this in the wild if you look at platforms like Codex or Edward unthanks sharing an unofficial Marketo MCP.

00:16:54: A model context protocol?

00:16:55: Yeah,

00:16:56: that protocol allows AI agents to securely connect to Marketo write custom messages and orchestrate dynamic email sends without a human ever opening the Marketo interface.

00:17:06: when The interface becomes the agent it just raises massive implications for the structure of marketing teams.

00:17:12: huge implications

00:17:13: I mean if the AI is doing the audience querying the campaign building, and that channel execution.

00:17:20: The role of the BDB marketer shifts entirely away from operational button pushing.

00:17:24: Let's actually pull that thread for a second.

00:17:26: if we zoom out to say twenty-twenty eight And this whole agentic vision is fully realized.

00:17:31: What does a marketing team?

00:17:33: Actually look like?

00:17:34: It's great question

00:17:36: because If the stitching has gone any execution as autonomous?

00:17:39: who gets fired in?

00:17:40: Who gets hired?

00:17:41: well it seems Like the traditional campaign manager roll basically vanishes

00:17:44: Right, and instead you need data governance engineers to maintain the lake house.

00:17:49: And probably behavioral psychologists or strategists who feed AI are right.

00:17:54: overarching business objectives.

00:17:56: That is an inevitable transition.

00:17:57: I think You move from a workforce of operators conductors.

00:18:03: The premium skill will no longer be knowing how to navigate the eccentricities of.

00:18:07: you know a specific software platform,

00:18:10: right?

00:18:10: It will be understanding customer psychology and possessing the analytical rigor to actually govern the AI's decision-making parameters.

00:18:18: So bringing all this together for the listener who has to go work tomorrow look at their own mark tech roadmap And somehow make sense of these shifts.

00:18:26: What is the ultimate takeaway from these past two weeks of industry upheaval?

00:18:30: well If we distill the mechanics of everything that we have discussed today, from Databricks to Liveramp and to the customer brain it really comes down this.

00:18:38: The future B-to-B MarTech is no longer about buying more disconnected tools for treat symptoms.

00:18:44: It's entirely about consolidating into unified zero copy data foundations Solving for deterministic identity natively And shifting your operations from manual campaign execution To autonomous agentic decision-making.

00:18:58: If you enjoyed this episode, new episodes drop every two weeks.

00:19:01: Also check out our other editions on account based marketing field marketing channel marketing AI in B to be marketing go-to market and social selling.

00:19:08: absolutely Check those out.

00:19:09: And before we let you go I want leave you with one final thought to mull over.

00:19:13: We've spent this entire deep dive unpacking how B-to-B marketing is rapidly moving toward being run by autonomous, reasoning AI agents.

00:19:21: But what happens in a few years when your enterprise buyers start using their own AI agents to research vendors read case studies and execute purchases?

00:19:29: Will the future of E-To-B simply be our marketing algorithms negotiating directly with their procurement algorithms in milliseconds something to think about as you review your data stack this week.

00:19:39: Thank You for joining us on This Deep Dive And Don't Forget To Subscribe!