How do you measure the real impact of marketing when every campaign unfolds across multiple channels, mixes digital with physical, and generates more data than most teams can handle? That's exactly what Abby Kearns and the team at Alembic are tackling. As Chief Technology Officer, Abby is helping build a marketing intelligence platform that doesn't stop at high-level trends or guesswork. Instead, it goes deeper to identify what actions actually drive business outcomes.
In this episode, Abby joins me to explain how Alembic brings together data science, custom-built neural networks, and lessons from pandemic-era contact tracing to solve one of the longest-standing problems in enterprise marketing. By ingesting data from CRM systems, media buys, social platforms, events, and even foot traffic, Alembic identifies causal relationships between brand activities and revenue. It's an approach that has already earned the trust of companies like Nvidia and Delta Airlines.
We also talk about how Alembic uses a combination of deep learning and large language models. While their custom algorithms surface insights and attribute outcomes, the AI layer makes the data digestible for decision-makers. The goal is to help enterprise teams move beyond anecdotal evidence and finally answer the question: what is working, and why?
Abby also opens up about the challenges that come with building and scaling a platform at the edge of what's technically possible. From managing massive data pipelines to keeping pace with generative AI innovation, the team at Alembic is pushing forward fast. This episode is packed with insights for business leaders who want to make smarter marketing decisions grounded in real data, not assumptions.
Are you confident that your marketing spend is delivering measurable value, or is it time to take a closer look at what the data is really telling you?
[00:00:05] What if you could finally answer the question, is our marketing actually working? Well today I'm joined by the CTO of Alembic, a company taking on one of the longest standing mysteries in marketing, True ROI Attribution. And they're solving it with a level of technological precision that's turning heads right across the Fortune 500.
[00:00:30] So my guest is going to walk us through today how Alembic is blending AI models originally used in pandemic contact tracing, pharma R&D and spiking neural networks to go beyond vanity metrics and uncover causality. Yep that's right, technology that helps you know exactly what that $1 spent in marketing actually did for your business.
[00:00:53] And backed by DreamWorks founder Jeffrey Katzenberg and NFL legend Joe Montana, Alembic isn't just building smarter analytics, they are rewriting the marketing playbook. So if you work in marketing and you've ever found yourself wondering whether that big campaign really did move the needle or just made a splash, this episode is for you. But enough rambling for me, it's time for me to officially introduce you to today's guest.
[00:01:22] So thank you for joining me on the podcast today Abby. Can you tell everyone listening a little about who you are and what you do? Thank you so much for having me. So I am Abby Kearns and I am the CTO here at Alembic Technologies. And I am responsible for partnering with Tomas Puj, our CEO and founder and the rest of the executive team on building and delivering an amazing product. And there's so much I want to talk with you about today.
[00:01:50] But before we do, for people hearing about Alembic for the first time, how would you describe it and the kind of problems that you're solving for your customers with technology? Top line story is we're a marketing intelligence platform. But, you know, for a more technical audience, what I tend to say is we're this amazingly complex data platform. We're ingesting tons of data from disparate sources, structured, semi-structured, unstructured data, tons of data.
[00:02:16] We're ingesting that data in and then we are leveraging a series of very novel, bleeding edge algorithms to surface insights, attribution, connect the dots to ROI, derive causality.
[00:02:32] All of that with the hope of really providing actionable insights for our customers, meaningful opportunities for them to really understand what is a very large set of data for the world's largest companies, if you think about their marketing data. And then for them to then be able to see that in a really nice, easy to understand format. I think you had me connect the results to ROI there because it's such a big focus on it right now and rightly so.
[00:03:01] And you've described Alembic's technology as almost the holy grail of marketing attribution. So how does your platform move beyond correlation to actually prove causality in that marketing span? Anything you can share around that or expand around that? Well, that's where the secret sauce comes in, obviously. So I think first, starting with a rich set of data and by rich set of data, what I mean is we're adjusting data from a whole lot of different sources.
[00:03:30] If you think about enterprise companies, and we're talking the biggest companies in the world. Think about how much data their marketing departments are sitting on. A ton. You know, everything from email campaigns and websites to outreach to conferences to sponsorships. Everything from arenas, for example, to billboard ads to think about advertising like Super Bowl ads to commercials.
[00:03:59] Think about everything from foot traffic into stores to what's happening on social media, YouTube, LinkedIn, TikTok, radio, TV, podcasts. It's just like the breadth of data. And so for us, the more data, the better. So we're able to ingest all of this data, both first-party data and third-party data. We ingest all that data in. And then that's really where we start the process, which is we ingest all of this disparate data in.
[00:04:29] We clean it, transform it, and then we apply time series to that. So we're really looking at seeing how this data behaves over time. And then that's when we're able to leverage our very novel algorithm, which is a technically, which is a spiking neural network, which allows us to look at that data over time. And then start to determine what are the things that are happening within that data that are interesting.
[00:04:56] So we're looking for, we're trying to be less of a dashboard and more what are the most important and interesting things happening. So we're looking for anomalies in that data. And then we're able to look at those anomalies over time and then start to derive causality, which means we're looking at activities that happen within a set of time. And then we derive causality.
[00:05:21] And then we're able, because we're also ingesting CRM data like Salesforce, we're also able to attribute those events to revenue. So we're really able to show our customers both what are the events that led to an outcome. So we can look at the steps that led to that outcome, and then we're able to attribute revenue to that outcome as well. So what happened that was interesting in my environment? What led to that event? And how much did it improve my business?
[00:05:50] Did it, how much revenue did it really contribute to my top line? And before you join me on the podcast today, I was doing a little research. One of the things that stood out to me was how statistical models from a pandemic era, from pandemic era contact tracing and AI used in pharmaceutical R&D, now being used and applied to marketing analytics. It's a great story, but can you tell me something about, tell me more about that for people who have not seen this?
[00:06:20] Well, one of the great things about our founder, Tomas, he really has a fantastic way of thinking outside the box. And so he had this epiphany when we were all becoming armchair data analytics people. And so, you know, we all became armchair data analytics people. And we're looking at the data and we're all deciding, okay, we're now all heavily invested in contact tracing, right? You remember when we were like, where did I go? Who did I see?
[00:06:48] And so a lot of really novel math came onto the scene around the same time. And Tomas had an epiphany. It's like, what's the difference between a contact trace? So identifying, you know, all the people you ran into that could have contributed to a COVID outbreak and a marketing attribution, which is what led to an outcome in marketing and mathematically nothing.
[00:07:11] And so that really began a multi-year effort to really understand how to take some of these novel algorithms that were created during the pandemic and start to apply them to very complicated data like marketing to help companies figure out just not only attribution, but what is driving the steps that lead to an outcome. What a great story.
[00:08:02] Well, it's a really big data problem. Like, for me, I've been in tech 25 years and most of my career, all of my career has been in enterprise, but most of my career has been on the infrastructure side. And so there's a lot of interesting correlations to a lot of the innovation that's happened over the last 10 to 15 years in infrastructure tooling. But the flip side of that is marketing teams are largely underserved.
[00:08:30] And not to say that there's not a lot of tech being thrown at marketing problems, but largely underserved in recognizing the breadth of the challenge, which is there is a lot of data. And I might argue that marketing departments and particularly in enterprise companies have the most data to sift through and figure out. And so really being able to help companies sift through all of that data and connect the dots is a really, really hard problem.
[00:09:01] And so I think it's been, you know, that's why we've been talking about this for decades, which is to say, how do I know what drove an outcome? And I think we all know now that it's never just the one thing. You don't get an email from someone and like, God, I got to buy that product right now. Let me buy it. No, you get an email and then you're like, you know, let me learn a little bit more about them.
[00:09:23] So maybe I'll go to YouTube and watch a video and then maybe I'll go to a conference they're putting on or read some of their blogs or really figure out how this solution works with the other solutions in my environment. And you really go on this, you know, for me personally, a meandering campaign. Maybe you also check Reddit. Maybe you look at what people are saying on LinkedIn and then but it's never just that one step that gets you there.
[00:09:51] And it's a combination oftentimes of digital, a digital journey and an offline journey. So maybe you read about this product. You read all the blogs. You watch all the YouTube videos. You're like, I really want to go to their conference. And so I want to go to a conference and learn more about it. And so you now have this digital experience, but also an in-person experience.
[00:10:14] And so where a lot of the prior tools really left off was the ability to connect the dots between what was happening in the digital realm, what was happening in offline, and how to bring those streams together to say, these are the big things that are really driving momentum in your organization.
[00:10:34] And the ability to connect the dots between those events, but to do it in a way that allows us to start to also move away from cookie tracking. So none of us really like to have cookie tracking, right? And we're all waiting patiently for the data that really goes away. And so one of the earliest aspects of the mission of a limbic was to be able to deliver all of that without understanding the user behavior that's specific.
[00:11:03] And so really looking at the population and the aggregate of the data and really starting to make a determination on, here are the things that happen. Here's the common patterns that a cohort of people follow. And here's how that impacts your company in terms of driving additional people into your stores or driving revenue.
[00:11:26] So what would you say is that makes a limbic custom built neural network different from, I don't know, off the shelf AI models, especially when handling the complexities of multi-channel marketing data? What is it that makes you guys stand out? Well, I think it's a combination of, you know, deep machine learning and AI. So we do leverage AI and specifically we're leveraging generative AI.
[00:11:51] We have a couple of large language models that we leverage, but we're leveraging those models largely, I like to say, largely to make things pretty. So we let our proprietary algorithms really carry the weight of identifying what is interesting that's happening, deriving causality and generating an outcome that we really want to surface for our customers.
[00:12:17] And then we're really just only leveraging the LLMs to summarize that information, make it approachable and, you know, really do even some persona based structure of that content. So that when our user logs into the system, what they're able to see is, hey, here's the top five things that were most impactful in my organization.
[00:12:40] Great. And it's easy, shows examples. And then I can click on through and I can read more about, oh, what are the steps that led to this outcome? Great. What are the specific campaigns? So I can keep kind of following breadcrumbs all the way through to go down to the lowest level if I want to. Like what specific podcast did someone listen to that really started this conversation and then follow that back up?
[00:13:05] But my journey starts at a really nice summarization of the opportunities for me. And then I can actually take action on that information or I can just copy and paste that and put that in an email to my executive team. So our goal is to really leverage where the innovation is with LLMs to derive approachability and ease of use for our customers,
[00:13:30] but also show them that they can follow that all the way through the system to get as much detail as they would like to get. And with companies like NVIDIA and Delta already on board, that should be enough to get everybody's attention listening. I'm curious, though, what types of insights are these enterprise clients discovering that were previously invisible? What was hiding in that smorgasbord of data?
[00:13:56] Well, fun fact, it's a lot of data, which I'm sure no one is going to be surprised about. But, you know, I think that if we look at the patterns and the use cases and the way each of those companies, among other customers, really leverage a tooling, it varies. Some of them are really looking for that summarization and their ability to bring this into their day-to-day workflow, which is, hey, I've got to really understand what's going on in my environment in near real time,
[00:14:23] which is a gap with existing technologies today or prior art technologies, might I say. So I really want to understand near real time, what's going on? What's driving meaningful impact in my business? What is what's happening with my brand? What's happening with momentum? Am I really starting to see outcomes from these investments I've made to understanding, did my large investment in, say, the Olympics drive value?
[00:14:53] And so really helping the world's largest companies figure out what is valuable to them and how do I really get a better feel for data? I think so I go back to where I've been as my career, which is I've spent most of my time, like I said, on the infrastructure tooling, which is everything from dev tools to automation. And I'm so used to having details at my fingertips. Data, what drove this?
[00:15:22] You know, we talk about tooling like observability and application performance monitoring. We're accustomed in running these large distributed systems to having a great amount of data, right? We know what went down, when it went down, what caused that. We can collect data on anything you want to know. But what was really surprising to me is that that wasn't the same case for marketing data. And this is, for many companies, their largest line item next to people is their spend on marketing.
[00:15:52] And what we really wanted to do is say, how do we give those users of these products, those customers, those insights? How do we allow them to have the same observability that I would have with a data dog with their environment? And so we really set out to say, how do we provide them as much detail as they would like to have on everything that's happening across their environment,
[00:16:17] but also make it really approachable for them to really start that journey to say, why does this matter to me? Why is this important? And then let me go find the details. And given the surge in marketing automation tools and AI-generated content, tech teams, CMOs, the C-suite, business leaders, they're all probably feeling a little bit overwhelmed with the sheer number of options out there at the moment. So how do you Olympic help companies make sense of what's actually working and why?
[00:16:46] Can you expand on that for me? Well, we really say, here's the things that are most meaningful. So we start with the biggest impacts. So we're really, again, not trying to be another dashboard, but instead say, here's the things that are really moving the needle. And then you can really start to unfold that at that point.
[00:17:06] You can say, oh my gosh, I had no idea that our CEO showing up on CNBC and talking about the work we were doing was going to drive so much traffic to our website, which was then going to drive people to our conference. Or I didn't realize that having our CTO on a podcast, maybe like tech talks daily,
[00:17:30] and was going to really drive a ton of interest in people wanting to learn more about our GPU capabilities or understand what are the options, best options for us to buy. And so really understanding how all of these things are interconnected is really the vision. And I think it's the most interesting part, which is when we look at data today, we've all been amassing data lakes for the better part of a decade.
[00:17:59] We have all of this data that we're hoping one day is going to prove valuable, right? And we're sitting on them and we're like, okay, I don't really want to know one piece of data. I don't really want to look at another dashboard. What I want to know is how is it all related? How do all of these things, how are they all connected? And then what are they doing for me as a business?
[00:18:23] So I really want to give our customers the ability to not only have that very detailed observability into what's going on, but then to also understand how all these things are interconnected. And can I then look at a string of events and say, oh, well, this is the path that a large population take. Let's double down on that. And maybe we don't invest in something else because we don't think it's driving as much impact.
[00:18:51] But I can start to repeat patterns and understand how customers really move the needle in terms of buying products, showing up at events, coming into our stores. How do I really want to steer that pattern and make sure that it's repeatable and scalable? But I'm also now able to do that with data versus I feel like this is what's happening. And so it's really helping our customers move beyond anecdote and into a data-driven conversation.
[00:19:23] It seems that your technology is solving so many big, big problems for some huge enterprises out there and household names. But I've got to ask, if we look towards you at the moment, what are the biggest challenges that you're focused on solving next? What keeps you awake at night, especially as Alembic expands into new industries and scales its platform? Because it's phenomenal what you're doing here and what you've achieved. But I don't want to make it sound too easy. There are a few challenges along the way, right?
[00:19:52] I mean, what doesn't keep me up at night at this point? Like the list of things that doesn't is probably shorter. But, you know, like any early stage startup, we're scaling. So we're scaling people. We're scaling our technology. We're really pushing also what the art of the possible is, as we think about the combination of both cutting edge machine learning algorithms, algorithms like spiking neural networks.
[00:20:19] But combining with, we leverage a couple of different large language models. And as anyone knows, it's even halfway tracking what's happening in generative AI at the moment. There's new innovation happening on that front on an almost daily basis. New models released, new benchmarks. And so really keeping an eye on those as well. And then really figuring out better, faster ways of ingesting, you know, ingesting data,
[00:20:46] transforming that data and really providing a powerful experience for our customers. And so all of that kicks me up at night. It sounds like you're going to be incredibly busy. And if you are kept awake at night, we do have an Amazon wishlist of books that we, the guests come on and leave behind for people listening to check out. And as someone that must read a lot, if you've got those sleepless nights, is that a book that you'd like to add to our Amazon wishlist or even a song to our Spotify playlist? What would you like to leave everyone listening and why?
[00:21:16] I will say that a book I come back to over and over again. Actually, I have two. If I think about the books that I recommend the most. First is Innovator's Dilemma by Clayton Christensen. An oldie but goodie, but I think anyone in tech should really read it and understand what that means and why just because you have product market fit today does not mean you will have that in the future. And then I would say the second book that I recommend the most is
[00:21:42] Move Fast, Break Shit, Burnout by Shannon Lucas and Tracy Lovejoy. And it's a really great way that for people like me that are always trying to really push the edge of what we can do in tech is to really figure out better ways to show up and do your work. Awesome choices. I will be getting those added straight to our Amazon wishlist.
[00:22:07] And for anybody else listening wanting to find out more about Alembic, everything we've talked about here, they've heard of names like NVIDIA and Delta Airlines being customers and so many more in the Fortune 500 in the pipeline. Want to find out more? Where would you like to point them? Please check out our website at getalimbic.com. And if you have any questions, you can either go there or you can find me directly on LinkedIn.
[00:22:34] Well, every day on this podcast, I always, one of the reasons I do it every day is I enjoy learning more about the technology that is solving real world problems. And for me today, the big takeaway is how at Alembic you're leveraging this sophisticated and complex range of technologies to accomplish what has long been elusive in the marketing world, but is now made possible through data and AI. So much more we could talk about on this.
[00:23:00] So I'd love to stay in touch with you, maybe invite you on towards the end of the year or early next year, see how things are evolving. But just thank you for joining me today and sharing your story. Thank you for having me. And I would love to come back. So a massive thank you to Abby for demystifying the technology behind Alembic and showing us that the elusive dream of marketing attribution is finally becoming a reality.
[00:23:24] What I find fascinating is Alembic is proving that data can do what your gut instinct never could. And by that, I mean quantify the impact of brand investments with scientific precision. And for marketers, this isn't just analytics. It's visibility, it's accountability and strategic clarity at a whole new level. That's your ROI on AI, my friends. It's a topic that I keep hearing about again and again and again.
[00:23:54] And that is why I just love inviting people like Abby onto the podcast. And I just get to learn so much. But over to you. I mean, I don't work. I'm not a marketing guy. I'm an IT guy at heart. So what do you think? Is causality the holy grail marketing has been waiting for? Have you faced similar ROI attribution challenges in your organization? What did you take away from our conversation with Abby today? Please slide into the DMs. I'll love to hear your thoughts.
[00:24:24] LinkedIn, X, Instagram, just at Neil C. I'm the easiest guy in the world to find. So please dive in. Let me know your thoughts. This is a dialogue, not a monologue. So I really enjoy speaking with you all. And if you are in the California area, I will be. I've got two events coming up in June. I'll be on the IT press tour in Silicon Valley and Cisco Live over in San Diego.
[00:24:49] So if you're in any of those areas, if you are attending any events, give me a shout. We can have a hot coffee or a cold beer. The choice is yours. But that's it for today. So thank you for listening as always. And I will speak with you all again very soon. Bye for now.