Katrin (00:01)
Welcome to Knowledge Distillation, where we explore the rise of the AI analyst. I'm your host, Katrin Ribant, CEO and founder of Ask-Y. This is episode 13. And we're sort of interrupting our series about agent e-commerce again to take a step back and look at how AI is changing the digital analytics industry. Maltech is at the core of what we use in digital analytics. Some of us are specialized in specific tools.
and build our careers on that expertise. The landscape is changing fast and I am so happy to have none other than Scott Brinker, the godfather of Mardic as my guest today. Scott, welcome.
Scott Brinker (00:46)
Thank you so
much. It's great to see you again.
Katrin (00:49)
you've been running Chief Martex ⁓
Scott Brinker (00:55)
⁓ we could dig much
further. So yeah, no, that's a very complimentary one.
Katrin (01:00)
You
have to stop at some point. You created the marketing technology landscape, super graphic, which is that graphic with all the logos that everybody knows about. basically, it's been the periodic table of Martech since 2011 now. And every year, you and Franz Riemersma put out the State of Martech report. The last one
was in December, if I remember.
Scott Brinker (01:30)
Yeah, we do two a year, one in the spring, sort of the state of Martech, and then we do in December always one that's sort of like, what is Martech looking like headed into the next year? So it was Martech for 2026, which by now we're in February, it's probably already way out of date.
Katrin (01:47)
certainly. And within all of that, you also spend eight years as a VP of platform ecosystem at HubSpot building out their partnership and integration ecosystem. And did I mention that you also hold degrees from Columbia and MIT Sloan? So with all of that, two questions to begin with. Can you tell us the origin? I'm sorry, I'm sure you've done this a thousand times, but please, can you tell us the origin story of the Martek landscape and ⁓
Tell us about what you are thinking about these days, like now versus what you were thinking then.
Scott Brinker (02:25)
Sure. All right. Well, you know, before I even like came up with that landscape, you know, I was, I was basically building a career at this intersection between the worlds of IT and marketing, you know, which you go back in time and these were two disciplines, two departments that were about as
opposite ends of the spectrum. I if you've gone to your like high school guidance counselor, you know, in like the 90s, you know, and you're like, I'm thinking about my grades are like, okay, well, it is over here. And you know, marketing is over here. ⁓ You know, and but of course, obviously, with the world going digital, it was faded. mean, these two worlds.
all sorts of wondrous things together. ⁓ But I was sort of very early on serving as an engineering background, but I love marketing. And so was sort of doing this like shuttle diplomacy between these two worlds. And one of my missions that I had was to try and help chief marketing officers, senior marketing leaders understand, hey, you really need to be thinking more about these technologies you're using and the operations for ⁓
And again, career marketing executives at that time were like, what? Why would I do that? I'm in marketing. And so I created that very first MarTech landscape kind of as if you were in a court case, this would be like, ⁓ let me introduce exhibit A to the case here of just showing them, like, look at all the different things, these different categories of these different products that may not be using all these, but odds are you're actually using a lot of them.
You know, and that these things have become so foundational to you being able to execute the strategy and the vision that you have, you know, and that was part of where the light bulb went off of like, oh, wow, I guess actually marketing does have a lot of technology in it. And that was fine. That was all I was hoping to do with that first one. You know, and there were like 150 technologies on that landscape, which at the time everyone, including me was like, oh my God, 150.
getting technologies, how will we ever keep track?
Katrin (04:41)
I very well,
I worked in an agency, I worked at Havas at the time, and I remember, literally I printed it out, because we're still doing that occasionally at the time, especially for the more senior people. And I distributed copies in meetings going, look, 150 tools, it's crazy.
Scott Brinker (05:05)
Insane and everyone's like, okay. Well, clearly this will consolidate You know, and so I I kept going back to it like year over year mostly because I was just one cup date like, okay Well, what's changed what's out there? and that was when this pattern of what essentially was exponential growth for a number of years it went from a hundred and fifty to 350 to then all of a sudden a thousand and then two thousand and four that you know, and this was certainly something I wouldn't have predicted and I guess
You know, now in the hindsight, we realized this actually wasn't unique to marketing at all. was, you know, the shift to moving into the cloud, the dynamics of, you know, so many ways in which the barriers tend to create software just kept dropping and dropping. mean, now, boy, there's not a category on the planet that isn't like flooded. You know, my tech is actually, turns out is one of the smaller categories on these. Like if you look at like, say the FinTech industry.
Katrin (05:53)
Yeah.
Scott Brinker (05:59)
It's something like four times as many products out there as, you know, Martek. ⁓ So anyways, was the origin story was like just trying to help persuade people to like realize. And I think even the landscape, it's become less useful over time because like really who can read the tiny little micro dots on any of this. The one story I think it's still useful for is it helps people recognize, okay, this is, this is a very large ecosystem. ⁓
things and it is constantly evolving and it's constantly changing and so if your narrative is like ⁓ I just need you know this one thing from this one vendor and I'm all set you're probably not looking closely
Katrin (06:45)
Yeah, that's for sure. And on your point about the ⁓ lowering of the barrier of entry to creating software, ⁓ at the time of recording, we are at the end of two weeks of what has been deemed as the SaaS apocalypse. Obviously, Cloud Code released 11 plugins, Cloud Cowork, sorry, released 11 plugins, and Wall Street went like, what? Everybody can now.
do all these things that all these big SaaS companies ⁓ rely on for margins, they're doomed, software is dead, ⁓ massive loss on the market, right? 300 billion of marketing capital evaporated from stock, I think in two days after the release. So ⁓ what do you think about that? Is software dead?
Scott Brinker (07:41)
Well, as someone who still holds a certain amount of HubSpot stock, I'm like, ouch, that hurt. I don't know. mean, you know, I'm in the camp and there's a number of other people who I think hold the same view is you got to look at a couple different things here. There's certainly the short term view of this is software isn't going anywhere. mean, like, you know,
Katrin (07:45)
You
Yes, I agree.
Scott Brinker (08:09)
all these businesses
marketing, like they all run on these foundational systems. know, your CRM, your automation system, you know, like that's not going to go away anytime soon. ⁓ And I also think this idea of, I don't buy into the ⁓ notion that people are going to like vibe code every piece of application, even moving further into the future. That being said, I think two things are true. I think
People are going to be doing a lot more vibe coding at a higher level. ⁓ They probably don't want to build all their own infrastructure, but they've got a whole bunch of ideas of their own special thing that they want to do. And I think companies that really lean into enabling that instead of trying to fight it, ⁓ I have an opportunity to be really strong platform plays in this next generation. But the other thing we have to acknowledge is, well,
I don't buy into the narrative that every mid-level manager is going to vibe code their own CRM. ⁓ The reality is, once again, the barriers to creating professional software have also just dropped and are continuing to drop more. And so the risk isn't like, yeah, my customers are just going to vibe code their own CRM. The risk is, there's actually going to be a thousand other companies that are trying to build professional CRM platforms for particular niches with particular ways.
And I think that's a real risk. so, you know, well, I think the market has overreacted in the short term. ⁓ I think we are, and I should probably just say, this is not financial advice. do whatever you do. You definitely don't want financial advice. But, you know, I think over time and it's, you know, it's weird because the other thing here is
Katrin (09:47)
Yes, let's disclaim. No, no, no, no financial advice.
Scott Brinker (10:02)
I was just writing about this earlier today. I actually think the future of software collectively is massive. I think we're going to see more software than ever. I think the challenge is it's going to be the companies that are sort of these large SaaS products today, the moats they've had and the business they've been able to defend.
Katrin (10:11)
I agree.
Scott Brinker (10:27)
it's changing. And that doesn't mean there isn't necessarily a way forward for them. I'm actually pretty optimistic there are ways forward for them, but it's going to be disruptive and there's a lot of uncertainty. And so while the market may have overreacted, I'm not sure the trend of like, hey, know, SaaS businesses as we knew them, probably going to look different over the next 10 years. I think that instinct's pretty.
Katrin (10:56)
That I think is absolutely accurate, right? I think the overreaction of the market and again, not financial advice, right? You remember when DeepSeek ⁓ released and that was the overreaction in the market with the Nvidia stock, mostly because very few people really understand the cost of inference. So they confuse the cost of trading and the cost of inference. ⁓ Today we see this reaction because
now everybody can vibe code software. Yes, five coding a little solution for yourself isn't exactly coding a complex enterprise piece of software, right? And I think what tends to be underestimated and, and, and this is something, you know, with data Rama was something that we, really had to like explain a lot to our investors and to the market is it's not because all of sudden something becomes.
way lower barrier to entry. Like for example, Google releases Data Studio and all of sudden you can do most of what we do for free. Yeah, well, sure you can, but it's very different to do a chunk for free, connect to an API and do a chunk for free and actually do enterprise-based reporting for your advertising accurately every day across all your stakeholders, et cetera. These are two completely different things.
I think this is sort of the same thing. I think that, yes, a lot of people are going to Vibecode solutions for themselves, and that's fantastic because that will heighten the expectation that people have of software can solve problems. And there will be more software used. And on the other hand, I just want to show you this.
Do you see my screen? Do you remember that? I still drink my coffee. So for the audience, if you're on audio only, I am showing Scott a very bad picture of a mug that Scott used to give out. Was this 10, 15 years ago? Something like that.
Scott Brinker (12:48)
⁓ I remember that.
Yeah, probably that. Yeah, at the Marcheca.
Yeah.
Katrin (13:12)
And the
mark says, so it shows basically a table with an executive, a meeting table with an executive, and then three snails ⁓ on the, Scott, maybe you just describe it. It's gonna be better.
Scott Brinker (13:25)
Sure. Yeah, so the executive, you know, round the conference table with these snails. But the executive is like just excitedly saying like, now that we've agreed on the marketing technologies we need, let's move quickly to integrate them into our operations. Which, yeah, I mean, I don't know, it's funnier when you see it, I think. yeah, I mean, the reality is, it's funny, this has always been my...
Katrin (13:49)
And then the tagline
says, technology changes exponentially, organizations change logarithmically.
Scott Brinker (13:57)
Yes, yeah. Well, I mean, this is the thing, you know.
Katrin (14:00)
I my coffee in this mug every day.
Scott Brinker (14:03)
I'm honored actually.
I mean this has always been the thing and it drives me nuts because I'm so known for that crazy martech landscape people assume that I am very like just pro like oh well you just buy some technology and then it's going to do everything you need. I mean it has always been the case that technology is the smallest piece of the equation of making these things work. It's what was it um Avanash Kaushik years ago right you know it's that it's like
10 % in the tools, 90 % in the people. To me, that still holds true.
Katrin (14:38)
It really does, no question. ⁓ So going back to the software aspect, you've outlined this idea of systems of truth and systems of context. Can you unpack that? Because I think it's really relevant to what we were just talking about, about the SAS apocalypse.
Scott Brinker (14:58)
Yeah, I mean, again, like for years we used to talk about systems of record and systems of engagement. And that's sort of where I see this as an evolution on that. The thing about systems of record is, turns out there's...
There's never one system of record. ⁓ mean, there's just like a tremendous amount of different sources of data that have relevance in different contexts. And so part of what you want to know is like, OK, well, where do I get the truth of a particular piece of data? It doesn't necessarily have to all be in the exact same place. But you have to know, what is the authoritative source? And how do I get to it?
Katrin (15:16)
No.
Scott Brinker (15:41)
You know, and so that's why I think about yeah, this sort of thing of like, you know systems of truth ⁓ I sometimes run to wonder it's a catchy label, but Truth is it starts to like get me thinking philosophically like well, what is truth? but alright, You know with the shift from like systems of ⁓ Engagement to systems of context is the systems of engagement were always framed that way because they're like, and this is where we engage with the customer
Katrin (15:59)
right
Scott Brinker (16:11)
Don't get me wrong, we still have these systems, we still engage with customers. But to me, what we're seeing more and more is this idea of like, OK, a lot of those engagement points, it's about how do we deliver the right context, like the right information, the right service, exactly what the customer wants there. But it turns out it's not just customer touch points that we want to do that with. It's employee touch points. It's like, ⁓ I'm looking to do this. I have to accomplish this piece of work. How do I?
What can I engage with that is going to make that as easy for me as possible? Like, OK, yeah, here's the relevant data for it. Here's the relative actions I want to be able to take on it. And this has exploded even more as we started talking about agents coming into the world of like, OK, you have all these agents. And one of the things I actually really like is this phrase of context engineering that people have started talking about with AI and AI agents, which my f**k.
oversimplified version. It's kind of an evolution of prompt engineering, where it's not just about like, okay, what's the set of instructions we give to an AI to do something, but it's like, oh, actually, we don't want to just give it instructions here. What we want to do is we want to give it access to the relevant data. You throw all the data at it, yeah, efficacy, cost, all sorts of things. So you really want to like, what's the contextually relevant data?
Are there particular tools that that agent is going to want to be able to use to like execute whatever you're asking us to do and to be able to package that up so that an agent as much as either similar as it would have been for like a human employee like how do they have things right where they need it or a customer who has a touch point how are they getting what they need I think of all of that is like this delivery of context and I'll also be the first to say I think that's really hard, you know ⁓
It's, you know, we're still at the mode where a lot of companies are still trying to just get these systems connected. ⁓ know, integration is by no means a solved problem in most, you know, tech stacks. ⁓ But just connecting the systems doesn't automatically deliver that context for you. I mean, like really being able to understand like how you curate in the moment in that particular context of what you're looking to do.
the right tools, the right action, the right instructions. That's challenging. That's why people today, they call it context engineering, because you actually have to do a lot of real engineering work to make that happen. But it's also where I think there's lot of opportunity.
Katrin (18:40)
in it.
It really is.
It really is. And I'm just going to do a shameless plug right now because it's just too relevant for me to pass this out. ⁓ This is exactly what is at the center of the thesis of Ask Why. Because when I start, obviously I've built analytics platforms before and I'm shockingly rebuilding an analytics platform. ⁓ And so I was thinking, okay, so what is the
central problem with analytics today. And I was like, it's really not in the tools. The tools are good enough today. You know, that wasn't the case 15 years ago, right? 15 years ago, if you're working with big data, the tools were not good enough. But then came the modern data stack and everything can be better, but honestly, they're good enough. The problem is the continuity of context between the steps in the analytics process.
Because when you do an analytics process, you have to connect to data, you have to bring it in, you have to organize it. You have to then model it for certain use cases to answer certain questions. And within that, basically what you are doing is you're taking a piece of reality, you are creating a model of that piece of reality, a data model. You are manipulating that model to get insights, answers, whatnot, truth, things.
And then you're re-translating that into reality. If at any point in time you lose the contact continuity in there, your answers are going to be irrelevant or false, right? And in marketing, the very, very basic version of that is, at least there's a dashboard to analyze spend and revenue gets in front of the market. So the market goes like, oh, revenue, where does that revenue come from? And now you have to retrace.
everything that figure has been through that probably has gone through three different people, five different systems, and follows the rules of one particular department and not the other. And that might be the right rules for this person and that might not be the right rules for this person. It's not like there's one set of rules that is right for every department. Finance recognizes revenue different from marketing and that's completely fine. It's different usage of the same variables.
And so creating an engine that will persist the context, not just the context, the technical context, but also the business context, the different places where all of these things exist, including the head of different people that at the moment where they're doing the analysis go, ⁓ actually, let's not forget the tagging was wrong in July and we got revenue doubled at that point.
We have to halve the revenue in this particular period of time. So keeping all of these traces and then creating exactly as you explained, creating the engine that is going to do the clustering and the ⁓ ability to have a retrieval that is relevant.
to the particular person and the particular circumstances in which they are prompting an alarm at this particular moment in time and inject the right context into it so that the results are relevant. That is a really, really, really hard challenge. It's really interesting, but it's a really hard challenge, right? But that is, I think, what is going to be needed in the future to make these disparate systems work because, as you said, there is no situation where we go into the one system of truth.
Scott Brinker (22:14)
Yes.
Katrin (22:31)
If that ever was possible with vibe coding, that is not happening. On the opposite, pieces of information that might not be deemed as institutionally important enough, but that are actually important at certain moment in time for certain people will exist increasingly in fragmented places. And so that content continuity is really, you know, in my opinion,
combined with natural language as an interface to all the tools that you use in analytics is really what will allow analysts to be able to answer the questions that they're asked to answer ⁓ better, you know, better, faster, or at all, quite frankly, because there's a point where at a certain level of fragmentation, it becomes impossible.
Scott Brinker (23:28)
Yeah, okay, well, I mean, yes, that example is like the perfect way of, yeah, really understanding like why this idea of like, okay, actually platforms and products.
that help us manage contacts, there's real value to be had there because it is incredibly complex. mean, you even just described it like, all right, just in the analytic side of it, I assure you, you take pretty much every other facet of business operations, and it's a very similar story. ⁓ And it's one of these things where...
Katrin (24:00)
absolutely.
Scott Brinker (24:05)
This is my advice to ⁓ software companies that are struggling with investor skittishness. ⁓
If you are going to make the assumption that the world is going to have an explosion of more software and AI and agents and all this than ever before, the question becomes is what you're offering as a product. Does it does its value increase.
as the number of things out there grows? Or does it shrink as the number of things out there grow? And I think, as exactly as you've described here with your platform, you're like, yeah, actually, the more distinct things you have, people like vibe coding out there and different contexts for different data on that, it becomes increasingly more valuable to be able to have a platform that can help pull that together, orchestrate that, govern that in a way that's efficient.
Katrin (24:40)
Yes.
Scott Brinker (25:03)
efficient, effective, reliable. ⁓ And anyway, so that's why I'm actually still very quite bullish on the opportunities for software platforms. ⁓
Katrin (25:11)
We too.
financial advice, but yes, me too.
Let's bring it back to, know, obviously knowledge destination as a podcast, we explore the rise of the AI analyst. So I kind of want to bring it back to the role of the AI analyst. If I'm an analyst listening to this today, what I'm hearing is basically my job is no longer building dashboards for humans to look at, really. I mean, it kind of is.
But my job really is to ⁓ become a context engineer, to understand how to actually use these tools in order to serve my goal, is ultimately answer the questions that people ask me to answer with data. Whether it's through dashboards or not, think it's completely whatever. That's actually irrelevant, I think. ⁓
That's what I see the main sort of shift in the analyst role. And for that, think that in terms of skills, if I were an analyst today in my, you know, 20s, 30s, and I had like 20, 30 years of career in front of me, I'd be thinking, so how do I need to upscale? What do I need to learn? And my personal opinion about that is you need to one, understand your tool really profoundly. You do need to understand how LLMs work.
If you don't understand the basics of how your tool works, it's not going to go great. Then you need to understand how to talk to the tool. And talking to the tool is really prompt and context engineering. And that's an art and a science. It's really something that takes practice and understanding of how context windows work, how attention and mechanisms work, ⁓ and how to kind of structure increasingly complex and long prompts.
in order to create an increasingly accurate result. The third one is for as much as we may not be writing code as much, we are going to be reading code increasingly. And reading code that you haven't written is a skill. It's really a skill. And so that's also something like, in terms of technical skills,
I'm certainly not of the school of you should not know how to code anymore. On the contrary, I'm of the school of you should know how to code and you should know how to understand how others code and how the machine codes because you're going to have to deconstruct that logic and put it together and because you're ultimately still responsible for that answer to the stakeholder. So when you started the Maltic landscape, you basically coined
the rise of the marketing technologies, right? That basically comes from that moment. What do you see today with the analysts versus at that time with the marketing technologies, similarities, differences, advice?
Scott Brinker (28:17)
Wow, yeah, there's a lot there.
I really like your point here about like knowing how to code is still actually incredibly valuable. You know, it's funny. So I started my life as a software engineer. And all fairness.
That's a pretty crappy software engineer. It's a good thing I moved on to other things professionally, but you know, it's probably been about 10 years or so since I've actually done any hands-on coding, you know, and then ⁓ like a few months back, ⁓ yeah, it's like, all right, well now let's take quad code and like, let's start doing stuff. And it's fascinating ⁓ just having enough of a perspective of like how this works. mean, not necessarily always knowing like, okay, what's the latest library on this?
⁓ The ability to both A, tell Cloud Code what I want in a way that increases the odds that it's going to give me what I want, but also then being able to, based on what it comes back with, because these things are very rarely one-shot ⁓ solutions, that's sort of an iterative dance, being able to understand enough.
of what it's doing and where is this going wrong and to be able to do that iteration back and forth, it actually lets you create really amazing stuff. I've seen this, a number of my friends who are really good engineers, ⁓ my goodness, the productivity they have with things like quad code is phenomenal because they understand so deeply what needs to be built and how it needs to be built that they can really masterfully have these things
doing the work, you know, to actually get stuff built, but then they also have the ability to provide like the judgment and the, you know, review. And so I totally see that expanding to, you know, other domains, which as you say, you know, from, you know, an analyst perspective. So I don't know, I mean, as a marketing technologist, I think the advice is very similar. Like you really do have to understand these technologies. And even though I'm a writer, I wish I could...
impart it to you through words alone. You have to be hands-on. This is a skill you learn through doing. ⁓ There's not enough YouTube videos in the world.
Katrin (30:30)
You too.
Scott Brinker (30:39)
⁓ But yeah, if you get a hands-on you do that and you keep doing it. mean, well, actually I see it, you know, some of the executives I admire most are those who still get actually very hands-on with these things. Not because you want your CEO, you know, being the one actually writing the code.
⁓ But because like if you see how actually knows how this stuff works and they have a sense of like their ability to have that influence their strategy the way they think about like this evolution to just understand the the the relationships ⁓ across things That's invaluable. And I mean, I'm not gonna name names or anything like this but I see it all the time like I I can talk to an executive and know like
All right, you've never actually put any hands on it. You have no idea what you're talking about. ⁓ They're usually the ones that are especially confident that they know what they're talking about. it's funny how that works. Versus those who like, yeah, I mean, they've actually embraced it and they've experimented. ⁓ And it's funny part of why they perhaps, it's not a lack of confidence, but they often then have a little bit more humility too, because it's like they actually have that deeper sense of like, okay,
I've actually personally felt how this is changing, the velocity of how it's changing, why, what that is like. It gives them a lot more empathy, I think, for their teams. ⁓ One of the things that drives me nuts is that sort of, we've gone through this over the past year of just these edicts from boards and say, use AI. ⁓
Katrin (32:14)
Yes.
Yeah,
Scott Brinker (32:21)
Generally not not helpful. Thank you. Thank
Katrin (32:21)
sure.
Scott Brinker (32:24)
you for that insightful advice ⁓ You know versus so yeah, they have enough like context and it's coming back to the word You know that they have the empathy for what their teams are doing they can give them direction they can get the feedback loop between those things and really appreciate it Yeah, so I don't know. mean, yeah to your original question. I don't think I answered is like you gotta lean into this stuff
Katrin (32:27)
Thanks
And I think it's a matter of grounding. I think that the problem when you stop being hands-on, and this is really something you have to manage in your career, right? When you start as a junior, you're very hands-on, you're a doer, and you execute things, and you become very familiar with the, most, at least anyway, like the actual operational aspects of what you're doing.
Scott Brinker (32:52)
Mm.
Katrin (33:19)
As you grow in your career, you have less and less time for that. You have less and less time to keep up with what's new. And I think from that perspective, this period is particularly challenging because the pace of change is, I mean, we've been through a ⁓ bunch of changes in our careers. The pace of change today is really faster than anything I have seen before. Faster and more profound. But if you do not...
put the time into having some hands-on experience, you won't be grounded in reality when you make decisions. It's not possible. I don't care how smart you are. It's just really not possible if you don't have, if you also, is something that I would say from experience, ⁓ having seen things like this before. When you have a period of change like this, where you are at the very beginning of something,
that is changing very fast and sort of trying to find itself. Nobody knows. So everything's changing and it's changing in a way that is small iterations and it's like...
Obviously the word comes in my head in French, which is not helpful. ⁓ You know, sort of like feeling in the dark, right? ⁓ If you miss the basics in that period, the entry ticket to understanding the basics of whatever that thing is, is become exponentially higher when you're three, four years into it and it becomes a more mature sort of corpus of knowledge.
We've seen that with ⁓ social media, we've seen that with obviously mobile, we've seen that with a number of things, right? So if you don't create the grounding at the very onset, it becomes much more expensive to do after. Would you agree with that?
Scott Brinker (35:17)
Yeah, no, mean absolutely because it's like ⁓ you just the
Will go back by the way and just acknowledge, you know, so when I left up spot I'm like, okay. Well, I'm gonna do this mar tech tops Analyst work full-time because there's just so much changing at least now if I'm doing it full-time. I'll be able to keep up Okay, that was a fantasy. I mean, I'm doing this stuff 24 7. I love it and I can't keep up I mean, you know, I mean literally like I ain't giving day just the flood, you know, and I'm not talking like oh There's this little thing over here and you know, mean
Significant things I've got this backlog of like major things that have changed. I'm like, alright, yeah, they have to get to that ⁓ So I think this is one moment where you just have to have enormous empathy if you feel like you're falling behind on everything ⁓ It you're not relative to you know, everyone else You know that thing like the the Lake Wobegon effect of like, you know, all the children are above average ⁓ it's almost like a
Katrin (36:07)
Yes.
Scott Brinker (36:23)
Inverse Lake Wabagoon effect here where everyone feels like yeah, I'm below average on this. Yeah. No, I don't get this It's it's like a really hard time doing that. But yeah, I mean Sometimes the way I phrase this is
The only way out is through. It's not gonna get any easier. The world is not going back to where it's been. The future is new. There is no way in hell you're gonna be able to master all of this, even keep track of all of it.
But that doesn't change the fact that if you do nothing, you are going to be left behind. You are going to become irrelevant to what has the potential to be really this incredibly exciting and transformative new age. ⁓ So you just have to start putting in the work. You would never guess given. But I exercise. And it's like one of these things of like, hey, listen.
I could do a lot more exercise, I could do a lot more bodybuilding, you know all this. ⁓ Not gonna happen. But actually just even like, know, doing like some runs, doing you know, a half hour to an hour, know, a certain number of days a week. Like, okay, that is by no means like this ⁓ expert paragon, you know, of the exercise world, but it's like, ⁓ you know, actually just doing that, like, know, healthier, I'm like happier, I'm like...
You you put in stuff not because you're going to be like the world's greatest, you know, at all these things, but because like doing a certain portion of this just becomes part of like how you live. And I don't know if this metaphor is actually holding up.
Katrin (38:14)
I think that's fabulous
advice. I really think that's fabulous advice because yes, it is a lot and is overwhelming absolutely for everyone, right? But consistency in effort, and it doesn't have to be an effort that is going to overwhelm you. Consistency in effort in just jumping at it a little bit every day and building. And for a long time, you will feel like you understand nothing.
and then it will start building. I agree the analogy with exercise is really spot on, think. I also exercise a little bit every day and I would not be able to do that much more. I'm at the age where that's kind of not really... But also at a certain age, there's only so much you can do.
Scott Brinker (39:06)
It's a little bit busy running a startup. I'm familiar with this.
Katrin (39:15)
But it does compound. It really does compound. it does compound and does give you that sense of grounding of at some point, yeah, I have these bases of knowledge. have these like, I'm starting to actually understand and words that I could not place in a relevant context three months ago. Now, when I hear the word, I actually understand what it is and I actually understand what this sentence means. mean, this really reminds me of the rise of ArtTech.
where we were all the kings and queens of acronyms. And my God, we had so many acronyms and they were all very bad. And when we started talking in between people who were in marketing, like we would understand each other. Absolutely nobody around would have any idea what we were talking about between the DSP, the supply side, the DMP, the et cetera. No one would understand anything.
And I think this is something, you know, but you just practice it enough and hear it enough, it starts meaning something.
Scott Brinker (40:21)
Yeah, no, it's a language. ⁓ Yeah, and it's a mental model too. I think that's what it comes down to. It's like, you we've had to update our mental model, you know, certain times, ⁓ you know, obviously computing, different mental model, internet, different mental model, mobile and so on. This one, I think you had said earlier, like I agree, it is both a much larger.
Katrin (40:35)
Yes.
Scott Brinker (40:48)
transformative change and it's also happening at a speed that's unprecedented. But yeah, it's basically another thing like, okay, you need to update your mental model to like, okay, this is what work is going to be like, this is what the world is going to be like, this is where I add value, what I can do, how can this augment me? you know, I mean, again, I always pause to acknowledge that's hard. I mean,
Change is hard. Most humans, ⁓ me included, sort of default reaction of like, I'd really rather not change. I was actually pretty comfortable where I was. you have to take, it takes effort to overcome that. And so I always am wary of coming across as somehow being flippant of like, well, listen, all you need to do is change your whole mental model and then you're good to go. It's like, no, that's really freaking hard.
Katrin (41:18)
It is.
Yes.
What's the problem with that?
So just to add a little layer to that, the tools are now starting to act on their own. And so in your latest Matic report, you've laid out a framework of three types of AI agents in marketing. Could you walk us through that?
Scott Brinker (42:05)
Sure. So obviously from the MarTech landscape, there's nothing I love better than categorizing little logos on a slide. Everybody needs a hobby. But as I was trying to think about, OK, how does this whole
Katrin (42:14)
Can I ask you,
you don't do that manually, do you?
Scott Brinker (42:21)
Well, not anymore. In the early days I did because I was a crazy person. But yeah, now it's all programmatic and yeah. But yeah, you know, actually the one for the agent, so I did do manually because for me it was even about producing a slide so much as it was my process of just sort of thinking through like, okay, what's, what is the hierarchy? What's the sort of structure of what's happening here? And it struck me that ⁓
Katrin (42:25)
Yeah.
Scott Brinker (42:50)
there felt like there were three very different kinds of agents. When we talk about agents of marketing, and to me the three buckets were, there's a set of agents that are for marketers. They're the things that basically help us do the tasks we're doing behind the scenes, whether it's like data analysis, helping to produce creative, like managing something in the internal workflow. Very useful. A whole bunch of actually really cool agents that have come out there.
Second kind of agent is what I would call agents for customers. And these are agents that are still actually controlled by the marketers or by the go-to-market teams. ⁓ But there are agents that are designed to interact with customers. mean, the most obvious example is these customer service ⁓ chat bots on websites, which, by the way, we can acknowledge. Those chat bots on websites have been around for years.
They've largely sucked. But in the past year or so, none of them have gotten really good because leveraging these LLMs, their conversational capabilities are much better. And we're getting smarter about connecting the right data sources to them. Again, they're still not perfect, but the resolution rates of customers being able to get the answer they want quickly, painlessly, that's actually pretty impressive. But anyway, so this is a thing that
Katrin (43:46)
Yes.
Yes.
Scott Brinker (44:15)
It's interesting because there are also agents like people talk about AISDRs and things like this where it's an agent that purports to help the customer, maybe can, but they're controlled by the marketer. It's still part of like, OK, this is my go-to-market. This is one of my channels of go-to-market is I'm running these agents to engage with customers. The third kind of agent is the one to me that is the most fascinating. And this is what I call agents of customers.
because they're not agents that are under any control of the marketer. They're agents that the customer is like, yeah. And the easiest examples of now of what people are doing is like ChatGPT and Claude and Jonah and stuff where they're saying like, hey, I want you to go out and I want you to like, I'm considering like switching CRMs. What are my choices? Can you like aggregate this data? Who's doing what? Set off deep research on it. It comes back and it's this information.
And by the way, you can go really crazy with this. I had that example that I think got me in some trouble with some people, where particularly in SaaS, often for SaaS at a certain time, one of the things that's actually hardest for buyers is pricing transparency, partly because these are complex products, but also partly because the sellers sort of leverage that as a way to move people through their funnel.
And I'm telling you, you turn to these agents on a deep research to be like, all right, what's really happening here in pricing? What should I expect to pay? You know what? How should I negotiate for getting the... It's kind of crazy. And so what's interesting is these are agents under the control of customers that are forcing changes in how marketing organizations...
are going to have to interact with the world. The other example, very common, is the shift from just Google classic Google search results and SEO to like, now everyone's using these AI assistants to sort of replace the way they think about finding information. So now all marketers are like, OK, I used to optimize for one agent. It was the Google bot. 20, 25 years, we got pretty good at optimizing for that one bot.
I've got this flood of other things. I'm like, yeah, how do I optimize to make sure that I'm serving that agent well so when that agent is answering something for a customer, I'm showing up where I should. It's these agents of customers that are the real disruption.
Katrin (46:53)
And yes, and those agents of customers, think one of the really huge disruption for marketers ⁓ recently is the agent with commerce, right? So chat, GPT with commerce, commerce protocol, et cetera. I actually have this theory that I like to test with you that this is going to be a shift in, it's basically going to kick off
a huge work stream in marketing departments. Because that is, in my opinion, comparable to the mobile platforming, because it's a shift in the customer experience, slash the shift to Amazon, because there's a loss of relationship with the customer. So it impacts your relationship with the customer, your loyalty programs, et cetera. And the GA4 migration, because really, when you think about it,
What needs to be done here is you have to optimize your website for two very distinct audiences, human and non-human, and not just the bots, the Google bots, bots that are actually going to actively going to look at very specific things in a very specific way that is not at all the way that the Google bots are looking at things. Commerce bots don't see JavaScript. It would be tremendously expensive if they did. It's not possible.
So now you have to basically restructure your entire website so that those types of bots can at least find the information. Then you have to get into the context window, an art and a science, right? And then from there, so because you've read on your website, you do need to redo your data layer. All your tracking has to change, right? You have to rebuild your data layer. You have to think about how are you going to measure this? You used to have a funnel.
Now for a large portion of your traffic, you will not have a funnel. You have all of sudden going to have a sales, you will not really know anything about that sales, not know anything of use for your optimizations about that client. And you also lose the ability to tell your story as a brand, to push the different cross-selling, upselling, et cetera. You now need to rethink about how you create a relationship with this customer.
if you are obviously in a category where it makes sense to have a relationship with that customer. So I can see sort of this massive work stream coming. think consultants are going to be very happy ⁓ because the agents with commerce, agents of customers with commerce are going to really disrupt marketing. What is your view on that?
Scott Brinker (49:47)
Yeah, no, I'm completely with that. I mean, that's another variation of the sort of agents of customers that, yeah, it changes the way in which businesses and customers are going to find each other, engage each other. And yeah, you're right. It's not just a Googlebot. It's like almost every facet of how we've orchestrated these digital interactions with customers and optimize them. Yeah, it's all up for
Katrin (49:52)
Yes.
Scott Brinker (50:14)
changing. ⁓ In fact, actually one that we don't talk enough about, I think is going to be huge is I think we are this close to seeing ⁓ an agentic revolution in email, where like the moment, you know, email agents get good enough that the customer is like, you know, my crazy inbox and all.
I'm just not paying any more attention to that. The agent is going to manage all of that for me and it's going to surface things up to me when I need it. And if I want it to go find something, it, it knows in the inbox who to go talk to and they'll go negotiate. And I mean, that's again, just another one where you're like, okay, for marketers, ⁓ you know, there's like, there's like a handful of things we've sort of like hung our hats on. One was like, Hey, SEO and organic search. got it. This is good. Now that's all being disrupted. Damn it.
Um, you know the other one like email like email marketing as a way of like, okay Well, we now have a relationship where relationship is defined like I have your email address Um, you know now, you know, this is how I use this as a channel and while the efficacy of email has slowly degraded over time because of all these you know bad actors, um Nothing like the sort of disruption that I think is going to be parallel to what we saw from, you seo to aeo
Katrin (51:17)
Yes?
Scott Brinker (51:35)
an email. But here's the thing, I'm optimistic that that's more of an opportunity than for the right kinds of marketers and the right kinds of businesses, you know, because at the end of the day, ⁓ marketers who even have good and valuable ways to engage their customers through email,
they suffer as a result of the fact that their email is still kind of lost in the sea of all this sort of crap. It's the same thing like SEO over time. Let's face it, mean, just the Google organic results, just they degraded over time. just wasn't, because it's not that there weren't good marketers using SEO tactics to do good content. It's that they had to compete with all these not so good marketers using those exact same tactics.
Katrin (52:06)
Yes.
Scott Brinker (52:27)
You know, and it's one of the reasons why the AEO stuff I actually am pretty bullish on too, because it's like, okay, if these AI engines, you know, the way they think and reason have a greater likelihood of surfacing like the true signal out of the noise, that those people who are truly providing the signal, that's actually, that's a win. And so I think we'll see the same thing in email. And it's one of reasons why I'm hopeful that, you know,
Katrin (52:49)
Yeah, that's true.
Scott Brinker (52:56)
You're concerned here of like, you know, if I no longer have the e-commerce experience that I've had with customers before, but I've got agents doing this stuff. Yeah. Is that just going to completely eliminate all these other options for, you know, upsells and cross-sells? I'm curious to see if that evolves because, you know, I don't know that for, I maybe if it's just about replacing, you know, toilet paper.
Yeah, I can really set that thing on autopilot. But I think for actually a lot of goods and a lot of services, I might want an agent to be helping me with things here, but there are other dimensions of the sales experience that even if I'm having an agent help me with it, I might want the agent to be actually paying attention to these things and helping me like, yeah, what is that cross-sell opportunity and why is that good? Okay, maybe the agent is providing a little bit of governance on my behalf in that, but.
If the agent actually understands more deeply what really matters to me and the agent is operating with the agency for me, not the seller, again, for sellers who are actually in a position to offer what's meaningful and relevant to me, the agent might actually be able to do a better job of surfacing that to me than I would have randomly found.
Katrin (54:15)
I imagine that obviously you get exposed to all sorts of new emerging technology. imagine like everybody who has a piece of technology reaches out to you because they want to be on the map. Right. So you get to see things that not everybody gets to see, whether they make it on the map or not. Do you see anything interesting coming up sort of in terms of
tools, technologies, concepts, ⁓ ways of addressing this agents of customer wave for marketers and analysts. That kind of like would address this notion of a certain degree of disintermediation of the relationship between the marketer and the customer.
Scott Brinker (54:59)
Yeah, I mean, I would say it's early. Like the one category where it's still early, but like, you know, things move fast. You know, there's this explosion of these, you know, AI engine optimization tools. I mean, there's hundreds of them, but there's actually like a couple dozen that have become like really fast growing, you know, successful companies. So I think that's a real example there. It's...
Part of the challenge is the landscape is so wide and so diverse. ⁓ just have to say I am by no means an expert in every single one of those categories. One category I pay attention to, and this is because coming from the role I did at HubSpot and even just my work for the Martech landscape overall, is I'm always fascinated by ecosystems. And one of the things I've just felt over the years is most companies do a really terrible job.
of activating their ecosystems and leveraging their ecosystems as a way to not just win customers but service customers. There's a lot we could do to make those things, you know, like much better win-win-win relationships. And there's been some interesting companies and steadily there's becoming more and more with AI that are about empowering ⁓ companies to do a much better job of treating the ecosystem as a core part of their go-to-market and their value proposition. Now,
For me, that's mostly been relevant in the tech industry. But what is very intriguing to me is this is not limited to tech. mean, almost any business in any industry, when you really sort of step back and see the threads, like, OK, for your customer, who else do they work with for what thing? You actually start to realize ecosystems are everywhere. We've just had a very poor ability, partly mental model, partly just a lack of tools.
to be able to really leverage ecosystems as a more significant part of how we build durable businesses. And to me, in some ways, because of the incredible noise with AI and all this explosion, I actually think ecosystems, the value you can have there, it's stronger than ever. so anyways, because I am passionate about that, I look at a lot of tools that are coming out of that space and I'm very optimistic.
Katrin (57:14)
Yes!
Yes, of course.
But that's fascinating because I can see how that would have a very different ability to execute with the ability to manage context that we have today versus before. I can totally see how that would be a place where there would be expansion, of course. But I kind of want to bring this back to the analyst persona.
working within an environment and organization. In the rise of the marketing technologies era, your premise was that marketing has to become a technology powered discipline and that marketing orgs needed technical people. And you said something that I really liked, which is it's not about finding the theoretical perfect tool. It's what you're going to do with it, which that is purely true. It's just
you know, that will never not be true. And that is where the real work is. And that's true, right? Obviously. Today, how do you see the marketing teams need to evolve? Obviously, as it comes to analytics, but as it comes to generally their need to have certain skill sets within the team, as opposed to outside of the team.
Scott Brinker (58:35)
Yeah, I mean, one of ways I've been thinking about this is grossly oversimplifying the work of marketing into three buckets. The creative and strategy, which is what people always think of as marketing, but in reality, most marketing games get to spend relatively little time on creative and strategy. There's the production and analysis work.
which is where the vast majority of actual hours have always gone into in marketing. And then there's that sort of marketing operations, marketing technology, like infrastructure foundation to support that. Which also it's been growing over time, but still relative to the grand scheme of the marketing board, a relatively small part. Because AI is absolutely changing the cost curve, the efficiency curve of that middle bucket.
⁓ like, you know, production analysis, it's just, it's becoming easier, faster, cheaper. ⁓ and it's the thing that's most disruptive to all these marketing orgs because I mean, again, it's not so much about saying like, ⁓ AI is just going to replace, you know, humans, you know, it's sort of like a one-to-one and wash, but it is the case that there's a bunch of work there that we needed humans to do.
that no, actually we can use AI and AI is really good and efficient at that. And so a lot of it then comes down to like, okay, now you've got this human talent that doesn't have to spend it, recropping images and Photoshop, like where do you redeploy that? And to me, this is where like leaning into on the strategy and creative side, man, you have the opportunity to do.
more experimentation, to be able to hone in on more niche opportunities because you have the wherewithal. It's actually ⁓ economically feasible to now run these more experiments, to go after more of these niches, to think bigger because, these ideas that before would have been out of your reach of time and budget to make happen, you're like, no, no, actually, we can do this. People who lean in that direction, I think, is phenomenal. And then I also think.
Katrin (1:00:20)
Yes.
Scott Brinker (1:00:37)
on the same side, like just the value of marketing ops and marketing tech to support all that. I mean, all this content engineering and all this stuff, it doesn't happen automatically, you know, and I think people who lean in that side of the space have enormous value to offer. I think my one feedback that I always give, you know, more senior executives, you know, if all you do is take the savings on production and analysis and just
put that to your bottom line. You're like, we need half as many marketers. Thank you very much. You know, isn't this great? I think that's an terribly short-sighted move because that efficiency gain has no differentiation whatsoever. Everybody's going to get the benefit of that efficiency. you know, so you have to really be thinking like, okay, maybe for those who are a little bit ahead of the curve, there's a little arbitrage opportunity at the moment, but
Katrin (1:01:24)
Yes.
Scott Brinker (1:01:34)
the speed at which this is moving that arbitrage opportunity is, you know, vanish pretty quickly. Question becomes like, okay, when production analysis is no longer any differentiator, yeah, how do you really think about changing the way your org leans into what it can do with creative or what it does from a technology perspective to tie this stuff together, you know, in a way that, yeah, doesn't happen automagically out of the box for people. ⁓
I get excited. I feel like there's so much opportunity for marketing, ⁓ but you have to know where to put it.
Katrin (1:02:00)
Right.
your point about efficiency, great, but you have to be in value creation. I couldn't agree more. How are you going to be sustainable as an individual or as an organization if you're not in value creation as well as inefficiency? It just doesn't work.
Scott Brinker (1:02:27)
Funny how it always comes back to that eventually.
Katrin (1:02:30)
It really does.
Well, Scott, this was really amazing. Thank you so much for talking to me. It was a fantastic conversation.
Scott Brinker (1:02:42)
Well, thank you so much for having me again. Yeah, I just love the chat, love getting your perspective on this. So thank you.
Katrin (1:02:49)
And
so shameless plug time. What do you want people to know from what you do? Where can they find your work? How can they reach out if they want to ⁓ propose their logo for your map, for example? I'm sorry.
Scott Brinker (1:03:06)
Yeah, ⁓ well, ⁓ actually recently just started a newsletter to you know, now that I'm doing this whole analyst thing full-time So if you go to chief martech.com and that's chief martech Without an H at the end All right, yeah, I'm not sure if I was a brilliant branding mover an absolutely boneheaded branding move I think it was actually the latter but ⁓
Katrin (1:03:06)
I couldn't resist.
We'll put the links in the show notes. Don't worry.
Scott Brinker (1:03:32)
Anyways, yeah, come to gdmardtac.com. It's all easy to see,
Katrin (1:03:37)
Wonderful! Well, we'll put the link in the show notes. Khan, thank you so much for doing this. It was amazing and I hope I get to re-invite you sometime soon.
Scott Brinker (1:03:46)
It would be wonderful. Thank you so much.
Katrin (1:03:49)
So that's it for episode 13 of Knowledge Distillation. If today's conversation made you want to experiment with AI for analytics, visit us at ask-y.ai and try Prism. Thanks for listening. And remember, bots won't win. AI analysts will.
