Best of LinkedIn: MarTech Insights CW 40/ 41

00:00:00: This episode is provided by Thomas Allgaier and Frennus based on the most relevant LinkedIn posts about MarTech in calendar weeks, forty and forty one.

00:00:08: Frennus is a B to B market research company helping enterprises gain the market, customer and competitive insights needed to drive growth and success.

00:00:17: And looking at what B to B leaders were discussing these last couple of weeks, you can really see a shift.

00:00:22: It feels like we're definitely moving past just talking about buying more tools.

00:00:27: Yeah, the focus seems different now.

00:00:29: It really is.

00:00:29: The conversations now are much more about, you know, actual system design, getting the data clean, properly integrating AI, and crucially how you actually orchestrate all of that to, well, drive revenue.

00:00:41: That

00:00:41: sets up our mission perfectly for this deep dive then.

00:00:43: We're going to unpack those key more tech insights for you.

00:00:46: zeroing in on what we saw is three big shifts in the sources.

00:00:50: First, building a smarter stack strategy.

00:00:53: Then, this kind of operational revolution that AI is driving through workflow redesign.

00:00:58: And finally, tackling that persistent headache, actually activating your customer data effectively.

00:01:04: Sounds good.

00:01:05: Let's jump in.

00:01:06: Okay, let's start with the stack.

00:01:07: It's where, let's be honest, marketers spend a lot of cycles.

00:01:11: And the big theme, maybe you so bring one, from the sources, is that Martek's success today, it's all about system design, not just collecting more tools.

00:01:22: Definitely not.

00:01:22: And there's a big reason why, inefficiency.

00:01:26: Dimitrios Kales shared a pretty shocking stat.

00:01:28: He noted that something like sixty percent of marketing tech just sits there unused in many companies.

00:01:33: Sixty percent?

00:01:34: That's incredible.

00:01:35: It's a

00:01:35: massive amount of waste.

00:01:36: And it's not just random bad luck.

00:01:38: The folks contributing agreed.

00:01:40: It usually comes down to the same problems, really systemic failures.

00:01:43: Like what?

00:01:44: Well, first, maybe no clear strategy when the tool was bought in the first place or, you know, really poor onboarding so the team never actually learns how to use it properly.

00:01:53: Or the tools just don't talk to each other.

00:01:54: Exactly.

00:01:55: Disconnected systems.

00:01:56: You buy something for one specific job, but the data it needs, it's stuck somewhere else in a totally different silo.

00:02:03: Okay, so if the way we're buying tech is broken, how do we fix the selection criteria?

00:02:07: You can't just go down a feature list anymore.

00:02:09: No, absolutely not.

00:02:11: And I'm around and also Vilsevon had some really solid advice here.

00:02:15: They were emphatic.

00:02:17: Choosing a stack has to start with defining the business outcome you actually want.

00:02:21: Not the features.

00:02:22: Not the features first.

00:02:24: Start with the problem you're trying to solve.

00:02:26: It's a specific result you need.

00:02:28: Otherwise, Anna pointed out, you're basically just buying complexity you don't need.

00:02:32: Makes sense.

00:02:33: So beyond that business outcome focus, what about the technical side?

00:02:38: What are the absolute must-haves for a stack that actually works?

00:02:42: AnimAround laid out three really crucial factors.

00:02:45: First, figure out the real total cost of ownership, the TCO.

00:02:49: That means thinking beyond the license

00:02:51: fee.

00:02:51: Training, integration.

00:02:52: Right.

00:02:53: Training, those inter-condition costs, ongoing maintenance, it all adds up.

00:02:56: Second, you need solid integration capabilities.

00:02:59: Preferably modern APIs because if your tools are fragmented, guess what?

00:03:02: Your customer experience will be two.

00:03:04: And the third.

00:03:05: Scalability.

00:03:06: But, and this is key, scalability that actually lines up with your realistic growth plans.

00:03:10: Not some, you know, massive capacity you'll pay for but never actually use.

00:03:14: That point about TCO is so important, especially when you look at older legacy systems.

00:03:20: John Miller had this quote that really hit home for me.

00:03:22: He described some of the older marketing automation platforms, MAPs.

00:03:26: He said, they're basically becoming marketing administration.

00:03:29: That's a great phrase.

00:03:30: It really captures it.

00:03:31: He compared them to these super complex old fashioned cocktails, like a brandy cruster or something that needs a specialist bartender just to make them.

00:03:39: They require constant, expensive workarounds just to function.

00:03:43: Miller actually put a number on it.

00:03:45: He estimates this drag costs the average enterprise about two million dollars a year.

00:03:51: He called it the marketing automation tax.

00:03:53: Wow.

00:03:54: Two million dollars.

00:03:55: Yeah.

00:03:56: Just in friction and workarounds.

00:03:58: That's the hidden penalty for not streamlining, isn't it?

00:04:00: It's huge.

00:04:01: So if that's the costly mess, what does the cleaner, more modern setup look like, practically speaking?

00:04:07: Tom Barth offered a neat framework.

00:04:09: He called it the real hat trick.

00:04:10: It's about getting three specific systems working together, like one single ecosystem.

00:04:17: Which three?

00:04:17: Your web experience layer.

00:04:18: So your DXP or CMS.

00:04:21: Then your marketing automation platform.

00:04:23: And finally, your CRM.

00:04:25: When those three are truly synced up, talking cleanly back and forth, that's when you unlock real personalization because the data flow is finally clean.

00:04:34: That unity seems key, getting away from what Franz Römerisma called the Frankenstack, that monster built from bits and pieces.

00:04:41: Exactly,

00:04:42: avoid the Frankenstack.

00:04:43: And Casper Kandelstorff added some thoughts on where the architecture is heading.

00:04:46: We're seeing sort of two big trends happening at once.

00:04:49: You've got the established big platforms embedding AI deeper and deeper into their core products, but you also have this... Explosion of composable, best-of-breed, often AI-focused niche tools popping up everywhere.

00:05:01: So which approach wins?

00:05:03: Well, the consensus seems to be that the winners will be those with a modular stack.

00:05:07: Architectures that are flexible enough to connect even swap out those best-of-breed tools effectively when needed.

00:05:14: Okay, so moving from the stack itself to how we actually use it, especially with AI, the message from the sources was crystal clear.

00:05:22: AI has to be baked into your operating model.

00:05:24: You can't just bolt it onto a broken process and expect magic.

00:05:27: Yeah, Zoey Merchant and Anthony Chooley really hammered this point home.

00:05:31: They stressed, you know, AI doesn't fix bad marketing or outdated methods.

00:05:36: It just makes the existing chaos.

00:05:38: Well, faster chaos.

00:05:39: Right.

00:05:39: You have to fix the underlying process first.

00:05:41: The leading companies aren't just like adding a chat GPT widget somewhere.

00:05:45: They're building an AI decisioning layer that sits above their existing stack, fundamentally changing how they make decisions at

00:05:52: scale.

00:05:53: That feels like a pretty big mental shift, though.

00:05:55: Yeah.

00:05:55: Moving beyond just asking an AI to do one single task.

00:05:58: It is.

00:05:59: Dale Bertrand emphasized this.

00:06:01: He said, The real productivity gains, the ones he measured at over fifty percent improvement.

00:06:07: They come when teams stop just doing one-off prompts in chat GPT and start building proper AI operations workflow.

00:06:14: Ah,

00:06:15: so building AI for systems, not just tasks.

00:06:17: Exactly.

00:06:18: And Dr.

00:06:19: Cecilia Dones framed this really nicely.

00:06:20: She called it a design problem.

00:06:23: She argued that historically our systems were optimized for, you know, output.

00:06:28: visibility.

00:06:29: I'm sorry.

00:06:29: But the systems we need now for the next decade, they need to be optimized for organizational learning and adaptation.

00:06:35: Elective intelligence basically.

00:06:36: That's the goal.

00:06:37: But there's a catch.

00:06:38: Arjun Pillai pointed out this big gap we have right now.

00:06:41: He termed it the agentic

00:06:43: gap.

00:06:43: Agentic gap.

00:06:44: Yeah.

00:06:44: Essentially, Martek is way behind sales tech and it helped in true agentic AI.

00:06:49: We're still often confusing basic API connectors, you know, like Zapier or Manion workflows with genuine agentic decision making.

00:06:55: Right.

00:06:56: So we're not quite there yet with giving the AI a goal and letting it figure out the steps and the tools only needing human checks at the end.

00:07:02: That real decision power isn't there.

00:07:05: Precisely.

00:07:06: that layer is often missing, and this internal gap.

00:07:10: It's becoming more urgent because of new external threats.

00:07:13: Julian Imchinsky flagged the rise of autonomous buy-or-side agents.

00:07:16: Buy-or-side agents, you mean AI buying things.

00:07:19: Essentially,

00:07:19: yeah.

00:07:20: AI that can research and evaluate vendors using all the public data out there, forums, reviews, analyst reports, without ever hitting your traditional marketing funnel.

00:07:28: Wow.

00:07:29: So they could decide not to talk to you without you even knowing they were looking.

00:07:32: Exactly.

00:07:33: It's a fundamental bypass of the sales and marketing signals we've always relied on.

00:07:37: Okay, so if the buyer side is getting that smart, our internal game, our data, our AI, it needs to be absolutely rock solid.

00:07:47: Evan Kubicek argued you need three foundational things locked down before you even think about reliable automation.

00:07:53: What are those less glamorous essentials?

00:07:55: Uh-huh.

00:07:56: Yeah, the unsexy but crucial stuff.

00:07:58: It's the ops foundation checklist.

00:07:59: First, a data dictionary.

00:08:01: Get everyone speaking the same language literally.

00:08:03: Stop having twelve different ways to write chief executive officer.

00:08:06: Standardization.

00:08:07: What else?

00:08:08: Second, minimum viable record standards.

00:08:12: Basically rules for data completeness.

00:08:14: You can't automate junk data.

00:08:16: And third, Living process documentation.

00:08:19: Not just how things are done, but why they're structured that

00:08:21: way.

00:08:22: Got it.

00:08:22: Dictionary standards documentation.

00:08:24: That foundational work is what makes AI dependable.

00:08:27: Yeah.

00:08:28: Justin Norris explained the technical why behind this need for structure.

00:08:32: He said AI reliability really comes down to two things.

00:08:35: Cask decomposition and owning the control flow.

00:08:38: task decomposition, meaning breaking big things down.

00:08:41: Exactly.

00:08:41: Break down a big complex task into tiny specific deterministic steps that an LLM can handle reliably.

00:08:48: And owning the control flow means your system architecture dictates the order of those steps, not the flexible and potentially unpredictable AI model.

00:08:55: Ah, so structure equals reliability.

00:08:57: Pretty much.

00:08:57: And this intense focus on architecture, on structure, it's completely changing the game for operations managers.

00:09:03: David Guardi made a bold prediction.

00:09:04: Oh, yeah.

00:09:05: He predicts that top MOPs, leaders who really master this stuff, who become, as you put it, context engineers, shaping data for AI across the whole GTM lifecycle, could see salaries hitting five hundred thousand dollars a year.

00:09:21: Half

00:09:21: a million dollars.

00:09:23: for marketing ops.

00:09:24: That's what he's predicting.

00:09:25: It's a pretty staggering forecast, isn't it?

00:09:27: It just shows how incredibly valuable that plumbing the structured, reliable data is becoming in the age of AI.

00:09:34: Okay.

00:09:35: That leads us perfectly into the final piece of the puzzle, actually activating all that lovely, clean, structured data.

00:09:42: Logan Woodbridge pointed out a really common bottleneck, especially in mid-market companies.

00:09:46: They invest heavily in data warehouses, right?

00:09:49: Centralizing everything.

00:09:50: But then they hit this glass ceiling because the marketing teams can't easily get the data out or activate segments from it.

00:09:56: So the data is there, but it's locked up.

00:09:58: Right.

00:09:58: And the result, the data team gets totally swamped with manual report requests, and the marketing team keeps missing opportunities.

00:10:04: because getting the segment they need takes too long.

00:10:07: This really shines a light on the critical role of the customer data platform, the CDP.

00:10:12: Yeah, the CDP comes up a lot.

00:10:14: How is it distinct from, say, a CRM.

00:10:17: Leon Hybin clarified that distinction really well.

00:10:19: He said, look, a CRM is mainly about registration, logging activities, managing contacts.

00:10:24: A CDP, though, its main job is integration, activation, and ensuring data quality.

00:10:30: It's about making the data reliable and ready to use across different channels

00:10:34: fast.

00:10:35: And it's not just one type of CDP, is it?

00:10:37: MaxCH mentioned there are different flavors.

00:10:39: No, definitely not monolithic.

00:10:41: MaxCH broke it down to four main types.

00:10:43: You've got data CDPs, analytics CDPs, engagement CDPs, and campaign CDPs.

00:10:49: And often, an effective strategy needs a mix of these.

00:10:53: Some focused purely on storing clean data, others built for real-time campaign stuff.

00:10:58: But even with the right CDP type, getting that real-time activation working is tough.

00:11:02: The sheer complexity of stitching tools together can kind of defeat the purpose, right?

00:11:06: Christian Tomas had a great example of this.

00:11:08: Yeah, he described this retailer using Salesforce Marketing Cloud.

00:11:12: To get the true, real-time, omnichannel personalization they wanted.

00:11:16: They needed four different Salesforce products working together.

00:11:20: Engagement, personalization, data cloud, and Tableau.

00:11:24: Plus, external agency help just to connect it all.

00:11:26: That level of architectural complexity creates so much friction in coordinating campaigns that, guess what, teams often just fall back to simpler batch and blast emails because it's easier.

00:11:37: So the architecture itself can sabotage the goal.

00:11:40: And while we're wrestling with tool coordination, we've still got this ticking clock on third-party cookies.

00:11:46: Jessica Beiffy shared a pretty alarming stat.

00:11:48: What

00:11:48: was it?

00:11:49: She warned that thirty-one percent, nearly a third of agency marketers are still completely reliant on third-party cookies.

00:11:55: That's just waiting for the revenue hit, especially with so many browsers already blocking them.

00:12:00: The sources were unanimous on this.

00:12:01: The only way forward is an immediate planned shift to collecting first-party data.

00:12:06: Through

00:12:06: direct user actions, engagement tracking.

00:12:08: Exactly, and server-side tracking methods.

00:12:10: That's the only sustainable path now.

00:12:13: Okay, so let's say you've got clean first-party data flowing, integrated via your CDP.

00:12:18: The final hurdle seems to be turning that data into actual decisions.

00:12:24: Oliver Yoan made an interesting point here.

00:12:26: Sometimes what looks like a data problem is actually something else.

00:12:30: Yeah, he argued that CMOs often think they have a data problem, like the data's bad or missing when really it's a prioritization problem.

00:12:38: Meaning the goal isn't just to pile up more and more data.

00:12:41: The goal is to make faster, better decisions that lead to ROI.

00:12:45: And Mike Rizzo added that if your analytics are misaligned of sales, marketing and support are all tracking different things.

00:12:51: Nobody

00:12:51: twists the numbers.

00:12:52: Exactly.

00:12:52: It erodes confidence and everything just stalls.

00:12:55: So often it's a communication breakdown, not a calculation error.

00:12:58: Theresa Madage offered a practical fix for that, using a simple framework for data storytelling when presenting insights.

00:13:04: Right.

00:13:05: She recommended that structure, context, conflict, insight, and then crucially, action.

00:13:10: Using that kind of narrative helps make sure the raw data actually leads to concrete decisions everyone agrees on.

00:13:16: It makes the insights usable for stakeholders.

00:13:18: Makes sense.

00:13:19: And just quickly, touching on a very specific area, lead gen and outbound.

00:13:23: There was some system innovation highlighted there too.

00:13:26: Michelle Laban talked about how agencies like Cold IQ are using something called master clay tables.

00:13:32: Yeah, these sound like highly structured ways to manage outbound data.

00:13:35: Handling block lists, enriching leads from multiple sources, apparently helping them triple their reach effectively.

00:13:41: And Manthan Patel mentioned platforms like Clodura that let SDRs do that whole complex workflow, finding leads, enriching the data, verifying it, exporting it all within minutes.

00:13:51: Whether you build these kinds of of systems internally or by a platform.

00:13:55: The underlying theme is consistent, systemizing that data flow, making it efficient and reliable.

00:14:01: That's the competitive advantage now.

00:14:03: If you enjoyed this deep dive, new episodes drop every two weeks.

00:14:06: Also check out our other editions on account-based marketing, field marketing, channel marketing, AI and BDB marketing, go-to-market and social selling.

00:14:14: And just before we sign off, thinking about all this complex.

00:14:19: system building we've covered, you know, the intricate stacks, the potentially high MOP salaries, the deep AI integrations.

00:14:25: we really need to touch on the security baseline.

00:14:28: Jubsho, George Jowski, flagged something pretty chilling that's emerging.

00:14:32: Oh, what's

00:14:33: that?

00:14:33: He called it spam GPT.

00:14:35: Apparently, it's being actively marketed and sold online as an AI-powered CRM for cyber criminals.

00:14:41: Seriously, a CRM for criminals.

00:14:43: Yeah.

00:14:44: It supposedly offers AI-generated fish and coffee.

00:14:47: analytics on campaign success, even deliverability testing, all designed specifically for running sophisticated scams.

00:14:55: quite frightening, a dark reflection of the tools we're building.

00:14:58: It really is, and it's a stark reminder, isn't it?

00:15:00: AI accelerates the bad actors just as effectively as it accelerates legitimate business.

00:15:05: It massively raises the stakes for security, for trust controls, and every single platform and system we deploy.

00:15:11: The architecture you build, it now has to be designed knowing there's an equally sophisticated adversary out there.

00:15:17: That's definitely a serious thought to keep in mind as you look at your own stack and processes.

00:15:22: If you want to stay ahead of the curve on the systems and strategies that matter, make sure you subscribe to The Deep Dive.

00:15:28: We'll see you next time.