Scaling Efficient Marketing Through Generative AI Automation with Lillian Pierson of Data-Mania

In this episode, we’re joined by Lillian Pierson, Fractional CMO and Growth Advisor of Data-Mania, to explore how AI is reshaping marketing strategy. Lillian shares a maturity model for adoption, use cases from across the industry and practical advice on aligning AI with business outcomes.

‍

Key Takeaways:

00:00 Introduction

02:53 AI shifts marketing from manual campaigns to real-time systems.

05:29 Marketing success with AI depends more on strong data foundations than flashy tools.

10:41 The bigger the AI investment, the more strategic de-risking it needs.

15:59 Humanic uses AI to find hidden user segments from behavioral data.

19:57 Marketing issues often stem from data silos and fragmentation.

22:22 Without a data layer, ad algorithms can’t optimize or validate market fit.

26:27 AI bridges marketing and tech through vibe coding and automation.

30:29 Try vibe coding tools — you’ll be surprised what you can build.

‍

Resources Mentioned:

‍

Data-Mania website
Cursor
Windsurf
Humanic
Growth Solutions

‍

Lillian: [00:00:00] we need to talk about like the different types of use of ai. and the amount of time and money it takes to build a really comprehensive, like, integrated strategy, 

‍

Anthony: Welcome back to Data Masters, the podcast where we delve into strategies, technologies, and leadership shaping our data-driven world. Today we're tackling a subject that's not just on the horizon, but actively shaping how businesses connect with customers and drive growth. The transformative impact of artificial intelligence on marketing, and I [00:01:00] couldn't be more excited to explore this with our distinguished guest, Lillian Pearson.

‍

Lillian is a leading voice in data science strategy, AI readiness, and enterprise transformation. As the founder of Data Mania, she's trained over 2 million professionals in data literacy, ai, and business leadership. Lillian Acts as a fractional CMO and growth consultant for a wide range of organizations, from startups to Fortune 500 companies.

‍

She helps them upskill teams and crucially align data strategy with tangible business goals with a particular knack for helping leaders build, as she puts it, unstoppable growth machines, often through data-driven marketing approaches. Her work, including highly regarded courses on platforms like LinkedIn Learning Bridge critical gaps between technical complexity and executive [00:02:00] level understanding.

‍

Given my own background as a CMO at companies like Qlik and Celonis, this is a topic which I'm particularly passionate about, and I know Lillian's Insights will be incredibly valuable. Lillian, welcome to Data Masters.

‍

Lillian: Thank you so much for having me, Anthony. I'm really excited for the conversation we're about to have. And this is a wonderful opportunity.

‍

Anthony: Yeah, so let's jump in. So. Let's start, you know, AI is clearly a significant topic, everything anyone's talking about these days.

‍

From your experience working with many different companies and seeing this through a variety of different lenses what are the ways and what are the biggest ways AI changing how marketing is being done, and how's it changing the expectations for marketing leaders?

‍

Lillian: Yeah, so there's two main things that I've been seeing in my work. And how AI is reshaping [00:03:00] marketing on a really day-to-day basis. One is there's a market shift from campaigns to systems. So where you used to do things like build like your entire launch kind of piecemeal and manually. Now most of the marketing is It's possible to be moving towards a systematic approach and solution to that where you're getting personalization at milestone levels and triggering messages in real time and just building relationships with customers. much more timely basis by using AI and systematic approach to marketing. And then the other part is the rise of the Vibe coder, I think is a really, really big deal. Because what that has done is it's changed who gets to build. so, as a person, you know, as a CMO, I understand, you know, most of the different parts of the marketing [00:04:00] processes, marketing systems.

‍

And so it's like you would use to have, to find a tool, an expensive tool to build to cover requirements. Or you would have to bring in quite a few team members. And now it's like, well, you can actually just. With, you know, your own two hands, build the tools, build what you need in order to get the job done.

‍

And that's powerful at scale. It's, so those are the two things that I think are, the biggest changes. And I would say what should be top of mind for marketing leaders is just to really. making sure that they're thinking strategically and also probably thinking a lot about the processes that are under their purview and start kind of breaking those apart and seeing where there's opportunity to create systems to create more efficient systems through ai.

‍

Anthony: I think this point you're making, I coding, CMO. There's, I think, a strong sense in marketing that some new piece of technology is gonna [00:05:00] be the thing that unlocks, you know, this new growth engine or the, you know, new increase in marketing qualified leads. You know, it's like we're always one step away from another piece of technology that's the.

‍

The great savior. And yet I feel like in order to really achieve our marketing objectives, it's often really about that solid underpinning of data. It's not just about some new piece of technology, but it's really the data foundation that's the key. So when you're talking to marketing leaders.

‍

About ai. I imagine that in many cases they're like, ai, it's gonna save the day. It's like our new savior. But how do you help them think about not just the tools and technologies, but that data layer underneath the data that, fuels that engine to really help them achieve their objectives?

‍

How do we link this to data?

‍

Lillian: Yeah, I mean, it's a great point. So, 1000000%. It's not about the tools, it's really about the business and how we can get the most efficient, [00:06:00] highest reward, lowest risk outcome for the business, given the resources we have. And so, honestly, it's not a lot different. What you're talking about with basically shiny object syndrome with marketing leaders.

‍

It happens in it, it happens all across the business. And as one of the things that, so I've been developing technical strategies for 20 years and have done, built a technical strategy for like organizations like the US Navy and have a lot of experience doing this. So, through those. Years and experience, I've developed a, strategic framework for building technical strategies.

‍

And so, we're talking about the marketing leader as the basic main stakeholder here. But in all honesty, your question points to the same solution, which is that we take a strategic approach. So the STAR framework is the framework I came up with and it, pretty, I think, simple for someone like you or anyone with. Strategic background. You know, you survey the industry, [00:07:00] you see what's out there, how other companies are achieving massive success in maybe what use cases you could even borrow from another vertical or something like that. But just really see, you know, make sure you know what's available with technologies. That are late breaking. And then, the tea is taking stock of your organization. So that's where you're looking at like, okay, so what are your data resources? What do we have available to us so we don't need to go out and buy new stuff? What data, resources what team members, what skillset, what technologies? And just trying to get, you know. Ahead of making sure we make the most use of the organizational assets we already have. Because that weighs in on the use case. It should be weighing in on the use case selection. And then the third phase of the framework is assessing your current states.

‍

So you're looking at like the value does basically looking at Doing a gap analysis across the different business units and then looking at risks [00:08:00] associated with different use cases and basically kind of trying to evaluate, use cases against one another to understand the full implications before selecting a use case because you don't want, A lot of times what I've seen, I mean I think you remember the Hadoop thing. So this is basically, this is the process that I walk through in order to. Really reduce risk associated with choosing any use case around which to build a strategy. So then the last step is forming the recommendations. But like, I think the use case selection is almost whole process. And then once you know you have, okay, you have the data in place, you have the skill sets or like if you don't. Habits. Okay, how much is it? What is the data partnership that might be available to you? How much is it gonna cost you to get these skillset sets you don't already have? What technologies you are gonna need to plug in that you don't have?

‍

And like really like thinking, okay, like I got this from engineering, so I'm must build systems engineer. yeah. And so it's the same with marketing though. That's exactly how you would have [00:09:00] to. to protect your investments, you have to think about data as a resource, and not just a resource, but like a, like capital.

‍

Okay.

‍

Anthony: couldn't agree more. And I think you know, this idea that well first of all, I think it's very helpful to have a framework to think about these things, and it gives people a structure in which to engage the conversation. and a little to your point AI is. Just another new technology.

‍

Of course, it's, you know, it is of course of the moment and it's the thing people are talking about. But I think the point you're making is we've been on the cusp of new technologies before and a structured approach for thinking about how to apply that technology into the business challenge we have is.

‍

You know, by its nature the way to engage it. It's not maybe to say it differently, it's not as though AI is totally different. It's just, to your point, another technology that we need to evaluate in the context of the goals we have or whether those are marketing goals or any other, is that fair?

‍

Lillian: Yeah, I would agree with that, but I would also say that we need to talk about like the [00:10:00] different types of use of ai. and the amount of time and money it takes to build a really comprehensive, like, integrated strategy, right? So, if you're just talking about using ai, like using Claw to generate marketing copy or something, there's not a lot of risk there. And so you wouldn't wanna build a huge strategy around that. So you could almost go ahead and do like, early adoption on certain things because, So long as you have thresholds, like you've already mapped out like what particular organizational risks there might be and like handed that off to the team member who's creating the content, whatever it may be. My point is that use of ai is very, very broad now. And so what I generally. Suggest is the more like money that's required, more time and money that's going into the investment of like the strategy supports this implementation. And if [00:11:00] it's a significant investment, then you need to do a lot of strategic like de-risking. but if it's just using ai like everyone else is using ai, then I would say it doesn't really matter.

‍

Anthony: Got it. let's move this from the theoretical to the realm of the practical. 'cause You've seen a lot of companies putting AI into practice here. And maybe if you don't mind sharing some of. Examples of. Where you see AI making a difference in what you were just saying you framed it as kind of almost like two different endpoints.

‍

Like, simple, easy things you can do that are low risk, high impact, and then sort of bigger, higher cost work that maybe requires a bit more thought. So, if you don't mind sharing, if you have some practical examples like that might help listeners and also think about how you think about the metrics.

‍

I, I suspect that the dimensions on which people measure performance change. When you're starting to use ai, you no longer need to take six weeks [00:12:00] to build marketing copy, to use your example. So helping people think through different. Metrics or data points they can use for tracking your performance.

‍

So, sorry, that was a lot to throw at you. So, but some practical examples and then how we think about measuring performance.

‍

Lillian: Sure. Yes. And so I basically am doing AI marketing all day, every day. So I've seen a lot of different. Ways in which AI is benefiting marketing operations. but when we talk about putting AI into practice, I think before getting into the details of like what that looks like, I think we need to get, kind of break it down a little bit of what that actually means, AI into practice. Because, so this is a little bit where I've been ideating on, kind of like making my own ip. So looking at like the maturity, basically making a maturity model for AI in marketing and kind of looking [00:13:00] at, trying to categorize how AI and practice really looks across an organization. So, how I see it, we've got like four main levels.

‍

You've got your individual AI usage, so that could be, you know, just the manual use of generative AI tools for productivity. So, saving money and maybe generating more revenue if you have a superior output. So you high have higher conversion rates, right? But, you know, it's just a team member using GPT or Claude or whatever it is a coder building, you know, using generative like co-pilot this sort of thing in order to get productivity wins.

‍

So that's a pretty low maturity. Use of AI at this point? it's not a lot of risk. Well, it depends on who you ask. You really have to be careful with IP for coding. But the next type of. Usage I see is like the team owned AI tool. So this is like the manual creation of tools through vibe [00:14:00] coding where you're able to save a lot of money because you're able to create efficiencies. Instead of having to work through three team members, you're able to build a. Tool that can work cross-functionally and get the output you need for a fraction of the cost. So, tools like that would be like cursor and windsurf, right? And so that's a more mid-level maturity type, categorization. And then you've got your ops level ai. So this is like using, yes. Okay. So we're still using generative ai, but we're now building cross-functional workflows. So we're building whole entire workflows that can be automated, not just. Tools that can do like one process, but now like how can we link processes?

‍

So, there's tools like you can build these automations yourself using like N eight N that's what I'm doing right now for and building a course on that. you can use tools like Humanic, which are basically, they've been built some, you know, Humanic is a SaaS company, so they've built. The workflow [00:15:00] automations for you and it's PLG growth product. so that's a mid to high level of maturity. And then you've got like the product level ai. So as someone who has a really like. Solid data background. this is traditionally what you would think of with an AI product, a machine learning, a deep learning product, computer vision. You're generating new revenue for customers or you're saving them money. And so an example of that would be like, I just off the top of my head, a company I've been working with is Growth Solutions where they've built a computer vision model and they're basically decreasing churn at gyms by engaging people at gyms using computer vision and machine learning, deep learning. But this is basically product level ai. Which is a significant investment compared to someone who's using AI on an individual level. So it's like, and it's a high maturity level. So it's, I think when we're talking about like what metrics to measure, we have to really look at like, how is AI being used?

‍

So, if we wanted to [00:16:00] talk about like. Level AI and Humanic. so Humanic works like you've got a SaaS product and users come in and they, you've got their. Behavioral data, their product analytics and on your website and then inside of the product itself. And so, they're leaving a data trail and then Humanic is doing micro cohorting, they're building segmentations and patterns. They're using the reasoning engine behind the, like, chat GPT, to identify segments that most people wouldn't be able to identify. Just by using the reasoning engine.

‍

And so then based on that once they have identified these segments there's a human, you know, human in the mix, right? So The marketing leader, whoever's using the tool has to agree. I think that makes sense. This segmentation makes sense. And then Humanic would go and basically execute a entire activation campaign with personalized [00:17:00] copy against each of these segments. And once it's generated the copy for each of the segments, it would go and a human in the loop get approval and then execute. And so basically what they're doing is. Increasing activation, decreasing churn doing that based on essentially like near real-time data inside of a SaaS product. And all of it is automated. So, if you're thinking about a use case like this, the type of metrics you would be tracking are very different than you would be tracking if you're just talking about like using chat, chat GT for creating launch content, right? So you would be more looking at like the activation rates of the users and the basically engagement and time to activation and retention. the general the SaaS activation and retention metrics that are standard across industry. So, when you're asking about the types of metrics that we should be [00:18:00] looking at, I think we really need to base that on Basically how AI is actually being used in practice and start thinking categorically about that so we can start then attributing metrics because, right now things are just getting so messy you know, with ai everything is ai.

‍

It's like we have to kind of, I think, break everything apart and start categorizing and really try to create systems strategic systems. [00:19:00] 

‍

Anthony: I love the example of thinking about using AI to automate a lot of the common marketing tasks. But it strikes me that one of the challenges associated with that is the underlying data becomes all the more critical.

‍

Now you've, yes, you're gonna have a human in the loop to approve copy, to think about approving campaigns, et cetera. But if the underlying data is a disaster, you're almost like. You know, making that person's job more difficult. Now they've got a, they're gonna have more automated campaigns that they need to approve and to put in place even if the underlying or because the underlying data is of poor quality.

‍

So how do you think about linking this question of these AI tools to the underlying data that the organization is using to drive, in this example marketing campaigns.

‍

Lillian: Yeah, it's a really great question. And I would say like, data is very [00:20:00] often a bottleneck in many of the companies that I have worked with. Not all, but and when you're talking about. cause we're really talking about data from like a data management perspective here, it's taking kind of a different lens to marketing. And so when I think about it, I can attribute when you're leading marketing campaigns it's easy to kind of. See a problem and think it's a marketing problem. But if you really look at it again it's actually a data problem. So the main data problems that I've seen come up over and over again is of course the data silos and fragmentation.

‍

So if you have no integration, you have no insights and that clearly not gonna work with many forms of ai. but there's also. Poor data quality issues. So especially if you're trying to do like a personalization use case, if you [00:21:00] don't have clean data pipelines and like, someone maintaining your data, and clear data governance policies, then it's really hard to execute. marketing campaigns? Well, in terms of like, say if you were to do something like PERS personalization, what I've actually seen be a bigger issue, really big issue that was unexpected. this is kind of, I guess this is a data fragmentation issue or. More than it is a lack of real time access data.

‍

But I remember working a project where we were trying to validate product market fit, and I came up with a product market fit hypothesis and we built the funnel on all this. So we just need a certain conversion rate in order to de-risk further investments. Right. To say, okay, like we have a market, we have a market fit for the problem that we're trying to solve, so great.

‍

That's gonna solve a [00:22:00] big problem for the company. And we go to like, run the a thousand dollars ads test, which is one of my favorite, lean test to get this done. And there's no attribution tracking. basically not just no attribution tracking, no conversion tracking. The data was just going. outer space. And so none of the algorithms, which are also machine learning, because the Facebook ads, Google Ads, they're all tracking conversions and so they can't optimize. So we can't get, any good results about any market fit because we can't get any conversion rates. 'cause the algorithms can't optimize on a. with no bottom to it. so it was like a whole problem and I didn't even expect that, you know, that's not something I would expect when I'm coming in to do testing for product market. Hit fit hypothesis is that like we need a data layer because otherwise you can't run ads to this in order to validate. [00:23:00] that's actually, I guess that's a data fragmentation. So that's the thing is like I'm really a marketing person now, but when I think about it, I think it's a data fragmentation issue or I don't know. I. 

‍

Anthony: it sounds like it's also a skills issue because what you're talking about is. Taking advantage of powerful AI technology that's only as good as the data that you can feed it. And in an example like that, you've presumably somebody in the system is not wired in the response rates so that the system can know whether the campaign is being effective.

‍

Lillian: no, it wasn't actually that. it wasn't the ads person, it wasn't the ads person at all. It wasn't the configuration. Uh, it was standard configuration didn't work because there was a problem with the website. There was no data layer. you don't necessarily need to have a data layer on most websites in order to just run Facebook ads or Google ads and get a conversion rate. So there was like more going on than [00:24:00] just that. And so you needed Yeah, maybe it was a skills. 'cause the second person we brought in built the data layer and so that solved the problem, but it wasn't easy and you needed developers and all of that. 

‍

Anthony: what's interesting about that and maybe to shift to this question of skills and sort of the human aspect of this and upskilling people as we see more ai. In use and in particular in use in marketing where maybe technology skills are not quite as prevalent.

‍

and you've done a lot of work in helping to sort of upskill people and educate people on the use of these technologies. maybe said very simply, what are you teaching them? Like, how do you get people to start understanding this stuff? Is it hands-on or is it more educational?

‍

How do you think about upskilling your clients and the people you work with so that they're more effective with these tools? 

‍

Lillian: yeah, so that's evolved over the years, which I'm really glad. My, my business has evolved and what I offer as a consultant has evolved. [00:25:00] So, you know, I have courses my most longstanding course is data science. So I'm teaching them how you know how to do data science and python, how to build recommendation engines, these sorts of things to the primary audience.

‍

There is data people, right? And then though more recently I've gotten into AI strategy and that has been awesome. But I'm really excited because a, over the years, like as I, I was working as a growth consultant for this entire time and had a data consulting practice, which is basically, you do not wanna have lines of business in a small business, just like, not ideal. And so I have gone solely down the growth. where I'm only serving as a growth advisor, a growth partner, CMO fractional CMO type of thing. So, that's a hard pivot. That's a, against the data science audience. And it seemed like, well this is a significant. Loss, maybe you could say I [00:26:00] guess. But actually now it worked out perfectly because AI came in and like really evolved over the years while I was evolving. And so now there's a huge opportunity here. So I'm actually working with LinkedIn and also with my clients who, like, when I'm coming in as a fractional, when I'm coming in as a growth advisor, I'm building them, their gpt, I'm helping.

‍

Training their marketing team on how to execute from the individual AI level, right? But now we get to go one step further. So now I'm teaching the technical audience how to use vibe coding tools, and also how to use tools like N eight N to build out automations. AI driven, generative AI driven automations to automate like really significant processes. And I use the marketing. The marketing function, all of the things I'm building are covering marketing because that's my subject matter expertise. But now in discussions with business side to start training [00:27:00] their marketers about vibe coding. it's pretty amazing that they kind of like the two things, kind of like dovetailed where I thought I kind of threw away a big part of my business, but actually it all came together.

‍

Anthony: Right. I think AI has this way of doing that and sort of bringing these two capabilities together. And this spirit actually, if you don't mind can you cast your eye forward and to the future especially from your vantage point? I'm. Really interested to know what you see coming down the pike in terms of AI, technology, AI capabilities, but also not so much the actual tech, but how people are using it.

‍

What do you see shifting and changing in the next few years? In, in terms of how people use this technology and maybe even, you know, what new technology is coming down that people should be on the lookout for.

‍

Lillian: Sure. yeah, so I think like what we have been talking about a. Lot. Just in this conversation is really what I would like label as legacy marketing [00:28:00] almost because it's is changing so quickly. So the problems that we have been discuss discussing, a lot of them are. they've been around a long time, but I think right now I would safely say that we're standing in a time like 20, 25, maybe the next five years to 2030, where we're looking at like. there's no label for it. Modern vibe, operations, vibe, ops or something. We're, We're building these tools like to automate our processes, automate big, significant processes of individuals work in order to create more efficiencies, right. And using AI agents and workflows.

‍

 but it's still. Requiring humans, right? It's still, especially the subject matter expertise. and you need humans like synchronizing between the different agents and optimizing and stuff like that. But I think in the next I'm hoping five years, I'm hoping that it's not any sooner than five years from now. [00:29:00] We're looking at like a future where these agents are actually, instead of having to have someone go in with, they're building themselves. They're adapting, auto adapting themselves. So you still need a human in there for, like systems level growth thinking, orchestrating multi-agent systems and interpreting the performance results and seeing where any changes need to be made to overarching. But it's much more of a systems based approach to marketing, like I talked about at the beginning. And I think that's when we're talking about like what technologies will come, I think that's what's coming. And I don't think it will take very long because these AI agent tools, development tools are like, they really emerged very quickly and so I wouldn't be surprised if the next, very soon we have agents that build agents if we don't have that already.

‍

Do you know, do we have that?

‍

Anthony: Yeah. certainly we think that a lot of the big AI companies are using their own models [00:30:00] to create derivative models. If you wanna call that agents. Training agents. I'm curious as well, if there were a listener to the podcast that was somewhat unfamiliar with this, or even, dabbling with using chat engines, what's a practical piece of advice that you might give a listener?

‍

What's something they can do tomorrow to help them understand this stuff and get more effective with AI in marketing?

‍

Lillian: yeah, that's a great question. I think I would definitely recommend trying out a vibe coding tool. they're free. Just windsurf cursor. Just get in there and ask it to build something that you need. And then maybe start thinking about what you're doing on a daily basis and start thinking in terms of standard operating procedures.

‍

And if you had to like build a system to like automate a certain line of a process. What would the steps be? This sort of thing. Because I think that if you haven't played with these tools to be very [00:31:00] surprised at what you can build and then if you can't build it using a vibe. Coding tool, you can almost certainly build it using something like N eight N. that's been my experience. So those things are all free and they don't even take a lot of time. And I think like. Just a few hours and you'll understand a whole perspective change on the future of marketing and also if you're a data professional or a developer.

‍

Uh, Just basically the future of knowledge work in general.

‍

Anthony: Yeah. No, I think that's great advice. I mean, it's like, just get your hands dirty, get in there, get using some stuff I really like this idea of think about a problem you have and then, build a quick solution to that problem. It is amazing what you can pull off with just a few minutes and some of the tools you described.

‍

Well, Lillian thank you for making the time and for joining us on Data Masters.

‍

Lillian: Thank you for having me, Anthony. It was fun. [00:32:00]