Best of LinkedIn: Martech Insights CW 38/ 39

00:00:00: Provided by Thomas Olkayer and Freeness, based on the most relevant LinkedIn posts about MarTech in calendar weeks thirty-eight and thirty nine.

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

00:00:17: You can find more info in the description

00:00:19: And we have A LOT TO COVER today.

00:00:21: honestly

00:00:22: Oh We really do.

00:00:23: I mean, if you're a strategic B-to-B marketing professional.

00:00:26: You've probably noticed just this massive influx of AI trends sweeping across your LinkedIn feed lately?

00:00:33: Yeah everybody is talking about it.

00:00:34: like every single post is about agentic AI or numar tech stacks.

00:00:39: Exactly so.

00:00:40: our mission for This Deep Dive Is to sort of cut through all that noise.

00:00:44: We're looking at why this rapid arrival of AI agents is Just totally colliding with broken marketing operations.

00:00:51: Right the operation's just aren't ready.

00:00:52: They really not.

00:00:54: And we're also good to explore why no one seems to actually own the tech stack anymore and how leading marketers are fixing their foundational data in measurement, to actually prepare for all this.

00:01:04: Okay let's unpack This.

00:01:05: yeah Let's start with that collision you mentioned because everyone is rushing to buy these AI agents right?

00:01:10: Oh totally it's The shiny new toy

00:01:12: It is, but they are quickly realizing that putting this blazing fast technology into these really slow outdated human operating models it just creates chaos.

00:01:22: It doesn't create efficiency at all

00:01:23: Which is wild because Efficiency Is the entire selling point

00:01:27: exactly and Greg Kilstrom shared This perfect example of this.

00:01:31: he looked At a specialty retailer That started using AI to cut down their campaign concepting time.

00:01:36: okay so like The brainstorming and initial drafting phase.

00:01:40: yeah

00:01:40: Exactly, and the initial numbers looked amazing.

00:01:43: I mean they cut concepting from eight days down to two.

00:01:46: Wow so a forex gain right out of the gate.

00:01:49: Right but here's the catch when you look at the total brief till live time.

00:01:53: So getting the campaign actually deployed to the market it only dropped from twenty six days To twenty four days.

00:02:01: wait really

00:02:02: yeah?

00:02:03: They saved six days up front But only two days overall.

00:02:06: Yeah

00:02:07: That's like buying a Ferrari for your morning commute, only to realize you're still stuck in the exact same traffic jam.

00:02:13: The car is faster!

00:02:15: But...the road is the

00:02:16: same.".

00:02:16: That IS exactly what happened… And Kielstrom explains why?

00:02:20: Because AI made generating all these different variants so cheap and easy that team just let it run wild.

00:02:26: Oh no.. Let me guess they made way too many.

00:02:29: I mean historically they produced about forty variants per campaign with the AI jumped into two hundred ten variants.

00:02:37: Yeah.

00:02:38: And the problem is every single one of those two hundred and ten variants still had to go through a manual promo terms review, you know for legal and brand compliance

00:02:47: right because You can't just have an AI hallucinating a discount code in pushing it live.

00:02:51: Exactly so.

00:02:52: those human reviewers were completely buried.

00:02:55: The review phase backed up from two days to seven days and their post deployment corrections tripled.

00:03:00: Wow

00:03:01: Yeah, the AI moved all of the work downstream but nobody actually documented a new workflow to handle that massive bottleneck.

00:03:07: Which brings up another huge danger with unmanaged A.I.. Melissa Rosenthal was posting about this recently.

00:03:13: You look at vendors like Demandbase and Adobe in Multiply.

00:03:17: They all just shipped A.i agents.

00:03:19: It's been a massive week for product launches

00:03:22: But Rosenthall points out because vendors are terrified if their A. I going rogue.

00:03:27: These agents are designed to wait for human approval before doing anything major, like shifting a budget or sending an email.

00:03:34: Which I mean makes sense you need those guardrails?

00:03:37: You definitely do.

00:03:38: but the vendors decided where those checkpoints go in the software and almost no B-to-B teams have actually assigned who on their end approves them.

00:03:46: Oh right!

00:03:47: Like is it demand gen.

00:03:48: Is It legal?

00:03:49: Exactly so.

00:03:50: you just have these AI agents sitting there waiting for someone to click approve And No one knows whose job.

00:03:57: But let me ask you this.

00:03:59: If performance agents are just optimizing for clicks, aren't we risking a world where all B-to-B campaigns look exactly the same because the AI just chases the safest metric?

00:04:09: Yeah that is a huge concern Because deciding to use more distinctive like brand focused ad even if it might have slightly lower click through rate initially That's a uniquely human decision

00:04:22: Right And AI doesn't know how to weigh long-term brand equity against a cheap CPC.

00:04:28: Exactly, usually a CMO or creative director has to make that call.

00:04:32: and what's fascinating here is despite all the hype about Autonomous AI it just not reality yet.

00:04:39: Harmonpreet core shared data showing that eighty point six percent of agentic workflows are still assist only

00:04:46: Over eighty percent.

00:04:47: Yeah,

00:04:48: meaning a human is still explicitly making the final call The AI as hybrid.

00:04:52: it's not fully native.

00:04:53: It can't run on its own.

00:04:55: Its basically super powered intern.

00:04:57: pretty much so.

00:04:58: think about your own operations.

00:05:00: What stands out to you?

00:05:01: About how your team reviews AI output?

00:05:04: do You have actual rules or are you just winging it when the machine hands you two hundred variations.

00:05:09: I bet most people who're just winking at.

00:05:11: And honestly, the reason these AI agents are getting so bogged down and bad processes is that marketing technology stacks have just grown incredibly complex.

00:05:20: Here's where it gets really interesting.

00:05:22: almost known as actually in charge of over-arching system.

00:05:25: Oh man!

00:05:26: The ownership vacuum.

00:05:27: It is wild.

00:05:29: Mike Rizzo shared this eye opening anecdote from the Unbound twenty six conference.

00:05:33: He was in a room with about forty different go to market professionals.

00:05:36: Okay So seasoned ops folks

00:05:38: Yeah senior people.

00:05:40: And he asked a simple question.

00:05:42: He asked, who runs your tech stack?

00:05:44: Like Who is the actual owner?

00:05:46: and

00:05:47: out of forty people only one could name an actual human being in their company.

00:05:53: You're kidding

00:05:54: One Out Of Forty

00:05:55: Just one.

00:05:56: everyone else just sort of pointed to different departments like oh sales runs The CRM and marketing runs the email tool but nobody owned the whole thing.

00:06:05: that Is terrifying.

00:06:07: By the way, if you are listening to this right now and realizing nobody owns your tech stack You should probably subscribe to this deep dive so we can help you fix it next time Because that is a huge blind spot.

00:06:18: It

00:06:18: really is, and Mark Long in Fabio Sanchez were discussing how this lack of ownership leads to what they call stack accumulation.

00:06:26: Stack accumulation?

00:06:27: Right I like the term

00:06:28: right because tools are just bought to solve immediate localized problems.

00:06:32: And then years later you have five different tools That i've met five different versions Of your customer

00:06:36: because They don't share any memory

00:06:37: exactly none of The data talks To each other

00:06:40: But surely people notice when these massive expensive systems stop working or aren't being used?

00:06:46: I mean, a CFO is gonna notice one hundred grand going out the door for nothing.

00:06:49: Right

00:06:50: you'd think so but Janelle McGrath shared a story that proves otherwise.

00:06:54: She talked about a twenty five million dollar manufacturer paying for this Massive enterprise Salesforce instance.

00:07:00: okay

00:07:01: So not cheap

00:07:01: no cheap at all, but essentially they were using none of it.

00:07:04: why

00:07:05: did a break?

00:07:06: No nobody called them meeting to abandon it or anything.

00:07:09: quietly stopped trusting it.

00:07:11: The pipeline stages in the CRM didn't match how their reps actually sold, so they just stopped logging things.

00:07:17: Ah!

00:07:18: So they kept a real pipeline and spreadsheets?

00:07:20: Exactly – shadow CRMs everywhere.

00:07:23: But McGraw's team came in moved them to HubSpot and configured it to actually match that process And the team adopted this in a week

00:07:31: Because software finally mirrored reality.

00:07:34: That makes total sense.

00:07:35: It is dangerous when you don't.

00:07:37: Amit Tiwari posted this horrifying example of hidden failures.

00:07:41: Oh, I saw this one!

00:07:42: Yeah

00:07:42: a leadership team had this revenue dashboard that used for everything and apparently there was the temporary filter left over from a CRM cleanup That hid an entire region From The Dashboard

00:07:53: Just wiped out An Entire Region's Revenue

00:07:56: Completely And it sat There For Two Whole Years.

00:08:01: Two Years They just got Used to the Numbers Looking Lower.

00:08:04: It wasn't until a brand new hire noticed it on day nine of their job that anyone realized.

00:08:09: That is just... wow!

00:08:11: But, it happens at the micro level too.

00:08:13: Jacob Mackie highlighted this disaster with just a forty dollar-a month calendaring tool.

00:08:18: Okay so something super basic.

00:08:19: Yeah

00:08:20: Super cheap but its shut down without warning one day And immediately broke five different workflows and three outbound webhooks

00:08:28: Because everything's wired together

00:08:29: Right.

00:08:30: The booking link wasnt' just calendar It was infrastructure, and it just goes to show that every single tool you plug into your stack becomes load-bearing.

00:08:39: Which means if our stacks are totally unowned in full of these hidden errors then the data we rely on to measure success is basically lying to us?

00:08:47: Absolutely!

00:08:49: So...to actually leverage the AI we talked about earlier We have to fundamentally fix how we measure value how we program our automation logic.

00:08:57: Yeah, We have to tear down the false idols of measurement.

00:09:01: Gillilouche posted a great takedown of influenced pipeline recently.

00:09:04: Oh

00:09:05: influence pipeline right?

00:09:06: The metric everyone loves to hate Right?

00:09:09: he analyzed A massive amount of data and found that a lead That cost two hundred dollars To acquire actually produced a cheaper final closed one opportunity than a lead that only costs forty Dollars.

00:09:20: wait really

00:09:21: yeah.

00:09:21: the two hundred dollar lead was ultimately cheaper.

00:09:24: yes

00:09:24: Yeah, which is the exact opposite of what an influenced pipeline metric or a cost-per-lead metric would suggest.

00:09:31: Right because if you're just looking at costs per lead You'd cut the two hundred dollar campaign immediately.

00:09:37: But wait isn't influence pipeline?

00:09:39: The exact metric every marketing leader uses to defend their budget and board meetings?

00:09:44: it Is And that's the problem

00:09:45: and this raises in important question Are we optimizing for cheap clicks at the expense of actual revenue?

00:09:53: Vikram C.L made a great point about this, he said knowing the cost of a click is completely useless if you lose the customer's identifier the second they land on your page

00:10:02: because You can't tie it back to the final sale

00:10:05: exactly.

00:10:06: Optimizing for cheap conversions just destroys lasting customer value.

00:10:10: But luckily there's a shift happening.

00:10:12: Roger Bahari Law noted at The Amazon ads partner awards that the industry has finally moving past basic ROAS.

00:10:19: return-on ad spend

00:10:20: okay Moving toward what?

00:10:22: proven incrementality.

00:10:24: Ah, proving that the ad actually caused the sale not just they happen to see it before

00:10:29: buying?

00:10:29: Exactly!

00:10:30: It's a much more mature way of measuring.

00:10:32: But to actually execute on that The automation itself has to change right To fix these rigid broken funnels we've been using.

00:10:41: I saw Matt Hines Patricia Diaz-Himes and Devin Rogolski all discussing this new platform called FAVE.

00:10:48: Oh yeah, Fave is super interesting because it's built by the co-founder of Marketo.

00:10:53: Right!

00:10:53: The guy who basically invented the old way of doing things.

00:10:56: Exactly.

00:10:57: and now he saying we are shifting completely from that era of rigid flowcharts

00:11:01: Because those flow charts just collapse under their own weight.

00:11:04: You try to map out a fourteen month buying cycle with Boolean logic And its just impossible.

00:11:09: It's a nightmare.

00:11:11: So instead of endless branching logic new systems like fave use objective based reasoning

00:11:17: Objective-based reasoning.

00:11:19: How does that work?

00:11:20: So you keep your rules around consent and frequency caps so you don't spam people, but then just give the engine a plain language.

00:11:28: objective like get this account to book a demo.

00:11:31: And the AI actually reasons its way toward that goal dynamically.

00:11:35: You're not telling it exactly which email to send on day three!

00:11:39: Just giving them goals in guardrails.

00:11:41: Right It decides next best action based on

00:11:45: real time data pulling all these threads together.

00:11:49: The future isn't about just buying more standalone tools, right?

00:11:53: Definitely not.

00:11:53: and as actually sewer warrants it's definitely not about relying on borrowed kagers or generic industry metrics to justify your budget.

00:12:01: no that won't work anymore.

00:12:03: dear Raj Kumar really hit the nail in the head.

00:12:05: he suggests that marketing teams need to start building what he calls a structured AI data center

00:12:10: A Structured AI Data Center so Making sure the AI actually has the context of your specific business.

00:12:17: Exactly, because AI without context is just going to confidently do the wrong thing very quickly.

00:12:23: Right

00:12:24: it needs to know.

00:12:24: you're historical sales data You real buyer signals.

00:12:28: otherwise It's just guessing.

00:12:30: and that takes us back To the ownership issue.

00:12:32: if your data is siloed Because no one owns the stack Your AI data center Is gonna be built on sand

00:12:38: Absolutely.

00:12:39: Which really brings us to a final, somewhat provocative thought for you to mull over today.

00:12:45: Bill Hobbit brought this up and it's fascinating!

00:12:48: Yeah his point really shifts the whole perspective

00:12:50: It does.

00:12:51: He basically said that if AI is making actual technology abundant and commoditized meaning everyone has access to same hyper smart agents then your competitive moat is no longer software you buy.

00:13:05: Right, the tech itself isn't a differentiator anymore.

00:13:07: Exactly!

00:13:08: The moat is your proprietary data it's deep customer understanding and actual human relationships you build.

00:13:15: Which kind of relief honestly?

00:13:17: It IS

00:13:18: But...it

00:13:19: forces big question What happens to marketing team when no longer have spend eighty percent time just operating machinery?

00:13:29: Who do become When Tech Just Works?

00:13:32: That is the question every marketer needs to be asking themselves right now.

00:13:36: Well, thank you so much for joining

00:13:56: us.