Pull Up a Chair: A Conversation with Lena Redko

Speaker A: Foreign.

Speaker B: Good morning, good afternoon or uh, good evening, wherever you are. So we are starting our podcast Analytics Friday. Today we have awesome guest. Lina is our guest. Lina, we're going to introduce yourself. Thank you for uh, accepting uh, my invitation to the show. M. It's a pleasure to have you here to uh, a conversation about marketing, analytics, AI and other things that we will talk about it. So nice to meet you. Thank you very much for your uh, time and let's start with the show Lina.

Speaker A: Awesome. Um, thank you so much for having me. Jorge. I'm um, pleased to be here. And um, I'm uh, Lena Radko. I'm based in San Francisco and um, I've spent um, most of my career in data analytics. So I'm excited to chat about it, especially during these transitional times, um, almost 18 plus years. Um, so looking forward to the discussion.

Speaker B: Okay, Linda, going to um, how you see the um. I see I'm going to give my opinion, but I think your opinion will be uh, most uh, welcome. Um, about uh, Marting AI. What do you think about this mix? Because um, there's a lot of going with Marting and AI. Not it's not so three years ago for now, now it's going to bump everything. But what do you think about this? Uh, and for the analytics like we are and uh, we talk about it, um, is the game changing or not?

Speaker A: It is, yeah. I do think that um, one can't argue otherwise. Um, it's definitely a game changer and um, you can see it on your LinkedIn feed and every conversation you are having at work. Um, the thing, um, that I've been thinking about a lot, um, you know what AI does is really democratizes how fast one can do analysis and you can do so much more and faster with fewer people. Um, one thing remains true though is that AI doesn't help with blind spots that the organization might have had. And so, um, for example, um, the insight remains, um, very human led. And what I mean by that is that um, even in the past, um, throughout my career, the key kind of thing, um, in insights and analytics was getting to the. So what? Um, so getting to the data and report, um, is one thing, but really interpreting that data and answering the question what do I do with it? And what is the right decision? I do think this still remains even with AI and democratization of analysis, still a human LED feature. Um, I think that intersection between data and strategy is really a fascinating space. Um, and really that context remains really important and even more important today than ever.

Speaker B: Yeah, um, so let me get this right. The AI is going to speed up everything, but we have to be some human in the loop. Because I noticed in the news that one agent was finishing all the databases, all the backups, all the recovery, uh, without, uh. So, um, what you think about, uh, not only the human in the loop that you said, that is, um. I agree totally. Um, and it's important to have human in the loop, to make that kind of questions that sometimes we take a long time on cleaning data, uh, and so on, and then to get the data right and to get the answers that we need. So, um, what, um, we have to put guardrails now in the AI otherwise we'll be stamped with no data. And um, the other side, that is okay. But with AI you can speed up, uh, for decision time. That is, um. It's a question of uh, the business nowadays. So the speed of the business, it's very fast. And we have to put speed in every side of our job too.

Speaker A: Yeah.

Speaker B: The question is, um, how you see, um, uh, taking more time for, um, formulation of hypothesis, uh, guiding the business more about strategical parts, construction data products as well. You can do it. What do you think about that? And the black side, that is, uh, attention. If you put everything in control of an agent, agent can wipe you off your data and you'll be bananas all the time.

Speaker A: Gosh, scary times. Yes. For somebody who's coming from the data background. You just described my nightmare. But um, I agree with you. Um, right. It's not um, just even the decision, uh, or judgment that comes with kind of like the so what and what you do. Um, it's also upfront. Right. Um, what is the right business question to ask? What are the set of hypotheses that you have? Um, it's interesting that um, early on in my career, like one of the first, um, places I worked at, um, was a B2B software company. And I, um, had a mentor there. And um, he was this charismatic smart guy, um, who basically told his team that you cannot present any analysis, um, unless you pick up the phone and call at least 20 customers. And the idea behind that is that you need context. Right? So, um, you are looking at the set of data without context. And it can mean so many different things. You really need that context to formulate your hypothesis. And unless you talk to somebody to, to tell you this is how I'm using my software, and maybe it's alongside your direct competitor and um, the data is kind of meaningless. So even now I think, um, the Risk is with um, this high velocity of um, automation, um, of analytics. You really need that time and context more and more upfront to make sure you're asking the right question and then in the end to make sure you applying the right judgment to make the decision.

Speaker B: Okay, Lina, just to sum it up, um, you're saying that I'm agreeing totally. Because you are saying very good things that I uh, think that uh, uh, a good data analyst, it's good if he knows the business, if he knows the context of how the data is created. I think that it's uh, the beginning of everything in my opinion. Okay. There is mathematicians and other things that the data scientists can do and I think they are awesome. I don't disagree with that, but I think it's for. To have business value you have to know the, you have to know the business where the business context, the business data. So in terms of uh, AI and AI applied to data engineering, um, uh, you have more time than ever to make all these things. Uh, if it was your priority to um, to have ah, a department or uh, uh, you are involved in the team of marketing, what was your um, priority in terms of adoption of AI?

Speaker A: Yeah, no, that's a great question and I think a lot of organizations are pondering on that. Um, what I'm seeing also in the market is a lot of companies are getting on the wagon as fast as they can. Maybe even without thinking through what the strategy is. Test and learn, um, is the focus. And to some extent I agree with that. Right? You kind of need to test and learn. Um, but at some point I think it's really important to put kind of like a strategic lens on it. And I definitely would not compare AI to kind of like how we approach vendor management. But to some extent, um, you know, you need to always start with the business question, right? Like what is it that we're trying to do? What are we trying to make faster, more efficient? And that's I think really helps the decision on where should we start. I think AI has amazing applications, is just really being intentional and strategic about it. Where to apply it in analytics space is really important. Um, for example, um, right. I think AI really um, is leveling the field in terms of like how fast you can produce analysis and research. But like I mentioned earlier, I think it also has a risk of amplifying the blind spots. So um, really being intentional in terms of like here's the process we want to automate is really, really important and kind of like being ruthless because there are vendors, uh, Popping out everywhere saying AI first. And it's really hard to believe, hard to like understand what you should believe. And so you can't test and learn every single vendor. So really being intentional in terms of what space you want to experiment on, where should you start first and go from there is very, very important. Otherwise you're going to start pouring money into like everything AI and in the end it's a bit of like garbage in, garbage out. So that intentionality is really really important.

Speaker B: Um, I agree Lina. Ah, let me ask you one thing that I um. It's always a thing that um. So if you have poor quality on data, we will be garbage in, garbage out, no question about it. But uh, in terms of um, business approach taking ah, and then okay, the tool, it's AI going to solve. Sometimes it's not AI it's going to solve, it's just an automation if then cycle and that's it. It's not worse than that. Um, but uh, everyone nowadays it's a big specialists everywhere bumping up as well.

Speaker A: Right?

Speaker B: Um, um, so it's difficult to see in the markets uh, with we're showing the way, the way that is correct. But you bring um, an awesome spot that is okay. We have two things here. Business, um, side. So we have to use AI to solve a problem that humans can cannot do. A pattern of data that we cannot see. Uh, uh, asking questions that we are not able to do it because the time to get the data prep and uh, data cleaning and so on. It takes so much time that you never get to the things to the answers or the CEO CMO don't have the time to market that decision because the decision is the opportunity is already, it's not there anymore. So the question is uh, about this is not the thing, it's not new. Uh, because now it's AI. We have different revolutions uh, across the markets. How about data literacy? Because everyone talks about data, everyone talks about AI but I think, I think it's not here in Europe or I think it's all world Occidental, uh, or not it's data literacy. And so what do you think about it? Because sometimes you are talking specific uh, metrics, specific API that has a vision in the context of business but the owner of the business does not understand what you are saying or we are saying to that kind of because they lack the data literacy. What do you think about it?

Speaker A: Yeah, no, again I agree with you entirely. Um, and I think in a way um, what all of these AI tools do is they Raise the floor. Um, you no longer need to be an SQL expert, you no longer need to be a data cleaning expert. You can generate an insight through AI copilot and Excel in minutes. Um, and so it gives you this false perception that you are in fact an analyst yourself. But I agree with you that that baseline understanding, right, like data literacy, as you refer to, is huge because you still need to kind of understand the basics. Otherwise, um, that appeasement that AI systems tend to do, right, like every time you run an insight and it tells you you're like the smartest person in the room, right. It's scary. Uh, your conclusion is usually. And the smartest if you, you know, like think of yourself as an analyst with no data literacy skills. So I think it's really, really important to, um, you know, keep that in mind. I think again, it raises the floor and it democratizes the analytics. But this layer of this, so what and what to do with it and judgment becomes even more and more important because you can make bad decisions really fast and it's very easy to have a spot in the data and the AI system won't notify you, hey, you might be missing a core component or data source. And so if you don't have anybody whose, um, expertise is rooted in business and analytics, you're going to totally miss it and make a very, very wrong decision. And again, being, um, in data for years, I can see how easy it is to kind of like miss a pattern and make a completely wrong decision about something very important. And you know, we talked about marketing effectiveness, which is quite a bit of art and science. But, um, you know, if you're missing like a core variable, you're going to put your, you know, $20 million in the wrong channel and actually like completely reverse your outcome. So it's just so easy to make mistakes and really, really important to have somebody with that experience, expertise who um, interprets the outputs.

Speaker B: Yeah. And as an example that in the ads about sneakers that the downsize of the sneakers was the sole, you know, that you put on the ground, um, was prioritized against the other, the face that you see in their shoes. So was, uh, after $50 million, uh, the guy stops the experiment because they are doing the A B test completely wrong. They put everything on the AI. I think like you said, it's very important to have human in the control to decide which, which experiment we're going to do and what is the outcomes based on the baseline that you have. Otherwise you'll miss the point. You'll miss the business. And the business will be burning money uh, for nothing, uh, just for nothing. Just to have that. So my question um, about it is. Okay, so we have to be careful about that. Um, but um, what you say to a CMO or cxo, um, trying to go to AI, we already talked some aspects but um, it's very important to have ah, people that are knowledgeable on the area that can give you soft guards and guardrails to do a proper job otherwise and agree on the metrics that will be accountable. Otherwise you'll be like everything has to go and everything. It's completely wrong.

Speaker A: Right, Exactly. As I was pondering about also um, how um today um, companies are still looking for people who know how to write SQL and Python even though AI can kind of do that for you. But I'm glad they are because at the end of the day um, you still need that data literacy aspect as the foundation. And yes the recommendation to the CMO still gets somebody who knows the, you know, the foundational aspects of it to just make sure your AI system is trained on the right inputs.

Speaker B: So in the future what you see that will be the. For the AI there's a, there's two sides of AI. I say um, to be so much automated that the human part will be diminished very McKinsey say more than two thirds of the business of the marketing department uh will be than by agents. Um but for other side we see here in Portugal and Europe as well lack of resources. So people. Everyone is going countries are getting very old on the average um, uh age. Um and um. There are very little resources to do the jobs. So in the part agents will be replacing people but in that part there are people that are not in the market. So if you don't get dead people in the market for aspect uh you can do more or you can say okay let's do a different thing. You can do more doing more marketing pieces marketing strategy or you can do the other way. That is many people are doing this way. That is a short. In my m. Opinion there's a short uh. Um. See because they, they only do for um. How you say it, uh for the short term. So they are cutting, cutting costs. So um, what is your um. You are golf half full or half empty? What is your, your opinion on that?

Speaker A: Yeah, it's, it's. It's interesting. Um, you know I do feel like at, at some point the pendulum might swing back a bit. Right. I don't know at which point though. But I do feel like at some Point. Yes, yes. Yeah. The key question is how far it's gonna go. Um, but I can already see, right, like if you're on LinkedIn, right. Um, I already am thirsty a bit for going back to like human, right, Human content and human layer. It's, it's too much and I think we're losing that human lens. And I'm not sure when the pendulum is going to go back, but I'm fairly sure that at some point there's going to be more need um, for that human layer, uh, to come back. And again, um, I do see a huge value and some things just need to be automated and there are a lot of tedious processes and analytics that really like no human should do. Just a funny story, my um, actually funny um, story that my entrance into the data world was quite unconventional. Um, my first job was this very tedious work of mapping um, customer records coming in in different languages to the right parent hierarchy at an Oracle database. Like just boring, tedious job and a perfect candidate for AI. Right. No argument there. Um, but I think it's just that understanding what is a good fit um, to automate with AI and what a human should do, I think that's the question still to be answered. And I hope um, there are not going to be too many devastating, um, things that will happen that will make this, this pendulum swing back. Um, and I do agree that there is a better use of human mind there than mapping data to the right hierarchy. But um, I think the time will just tell when exactly the pendulum will go back. But I'm looking forward to it as well because I think AI cannot fully replace the judgment.

Speaker B: I think, um, I completely agree with you because there are some tasks that are very painful to the human side and I think um, with a human loop on the loop that can be automated in some way. Uh, but the thinking, the strategy, the auto context that with the data that we already have and the context we are living because we are selling to humans, not for machines yet, but we have another thing that is um, agentic E commerce that you go to the generative interface to the chatbot and say please, um, see my toaster. Um, that better fits my profile for costs and for functionalities or features, if you want to say that. So, um, I agree in some extent, but um, in the other way, uh, what is your opinion about um, uh, the human panel goals goes to the human in the loop. Again, not too much automated. I agree. And there's uh, things about the geopolitics as well that are going so even the war, it's going to be AI AI first, not human first. And there's a problem there as well. Um, so that is ethical. Ah, so Hawaii is going to be ethical working with humans. So that is the question that we have to answer now because it scales too much. Um, there'll be. The human will be not part of the chain and it's quite scary on that point. Right.

Speaker A: Um, I agree. I think we're um, you know, um, I hope maybe in some countries the leadership will step up and take that role of um, ethical AI. I um, wish it was happening more here in the US and it is quite scary. And um, one angle sure was E commerce and agent E commerce. That's not maybe hugely devastating circumstances but with other aspects that you bring up it's. It's definitely the fallout can be much worse. Um, I think it all has a huge impact in the US I see quite a bit that the marketing is shifting. Um, we are trying to market to AI agents now, not to people. And there are a ton of startups actually popping up in this field as well. Um, how do you capture the attention of an agent versus a human? Um, the implications of all of it are huge, huge. And again lack of guardrails is scary these days and I am hoping that maybe there are some um, parts of the world where the leadership can be a little bit more um, put their foot down and establish their guardrails where the rest of the world can adopt them.

Speaker B: Um, I'm sure of that because um, here in Europe they are doing. Sometimes they are critical about so much guardrails and so much bureaucracy to fulfill in some aspects. But then you see some parts of the world um, they are not guardless of anything. So everything is possible, everything is funded, uh, get funds, public funds as well. But the way that um, we should have a balance of this otherwise um, the human part will be completely finished. And um. And maybe you are not talking anymore in the future because uh, we'll be machines to machines or something like that. I hope not. You're going to go to this area. I think human in the loop is very important to power the people and to be um. There's uh, an interest part that I would like your opinion as well. It's uh, how you differentiate because LLMs are so intelligent if uh, they are not um, um elevated in some way by the human side and to give different aspects and blind spots and strategy. The only way to differentiate between a typical aliarm prompt and answer to the human that wants to be or the brand that wants to be different for selling to humans. Right. What's your say about it?

Speaker A: Yeah, no, I, I think the differentiation of products you mean. Right, like how do you differentiate in the market?

Speaker B: Um, in the same market.

Speaker A: Yeah, yeah. Um, you know I, it, it's a, it's a great question because I think the marketers used to be right. Like it, it's, it used to be more of art than science and I think today the pendulum already kind of like SW and even in marketing I feel like it's going to swing a bit back because it's all about data. And being a data analyst I actually um, I should love it but I actually don't because I think sometimes we over complicate things. Right. And marketing becomes more about like we measure every single interaction. Um, meaningful or not. There's always some kind of like data analysis aspect to it. Um, but I think um, really like the campaigns or the marketing or the products we remember and really stick in our brains, um, are really coming more from the creativity side. Right. So um, um, the number of clicks on your TikTok video is not going to give you an opportunity to come up with a groundbreaking product. Um, it's really something else. Really understanding in depth what the consumer wants or even expecting that. Um, and I think that differentiation aspect comes from that really is really understanding the consumer really, really well. And I think we're going a bit further and further away from that. It's more about maybe signals that are not as important because at the end of the day it's the consumer you're trying to win, not a machine. So I think if you are trying to differentiate, try to go back to what a person wants, how they make decisions. A lot of them, um, you know, a lot of those signals do not sit in the behavioral data. Um, right. There is a reason why research and analytics is such a powerful combination is because it's as much about how people think and feel as well as what they do. And I think that combination is something that makes products very powerful. Right. Is like being able to combine kind of like this subconscious with conscious is I think is a way for companies to really differentiate and come up with amazing products versus being just behavioral data driven, collecting every signal and ultimately learning from AI. So that's what my take to it. I think there's a little bit going back to the roots of kind of like marketing and understanding the consumer.

Speaker B: I think you are on the spot, on the sweet spot. Let's go back. Let's use data, enough data but let's go to the people that are the consumers. I agree with that. I like a lot of data, a lot of technology, but uh, I like more people. It's more important than the others. So data and AI will be good to help us, um, working on that goal, on that objective, not the way around. So, uh, Lina, we are almost done for this episode and um, thank you very much. I would like you to you to give a recommendation for all these. To have a. I don't know if you have a book recommendation there to. Or a movie.

Speaker A: I do, yes.

Speaker B: Or something. Please let us know what you are reading, what you are recommending to our audience, please.

Speaker A: Yes. Um, you know, there is um, this favorite book of all time for me and I've read it several times and I keep coming back to it. Um, and it's called Lust for Life. Um, it's about Van Gogh, written by Irving Stone. It's an amazing, amazing book and if you are into kind of arts and impressionism, um, is just such a wonderful book. And um, back in the day I actually followed kind of like the book locations and traveled to different spots and so um, you wouldn't be able to put it down. So that's my favorite book and hope you get to read it.

Speaker B: Thank you very much. I will be in our episode on the recommendation because we have all these recommendations for our audience, our guests because uh, it's important to give different uh, sides of the personality of our guests too. And it's awesome to get yours as well, Lina. So thank you very much for your time and attention. It's a great conversation about marketing, AI, analytics people. Everything was very good. Thank you very much for your time and um, you see you next time.

Speaker A: Thank you so much, Jorge. Pleasure to be here.

Speaker B: Okay, bye bye. See you next episode. Analytics Friday.

Speaker A: Mhm.

Speaker B: Podcast.