Hey, everyone. Welcome to More Numbers Live, the data show focused on turning data teams from support functions into engines of growth. I'm Ollie Hughes. I'm one of the cofounders of Cal dot co. We're a Canvas based BI tool. We're all about turning data teams from dashboard factories into their organization's problem solvers. Each week, I have the joy of interviewing another data leader who is driving massive value into their organization, and this week is no different. This week, I've got Lina Mikolajczyk. She's the head of analytics at Bumble. She's an absolute badass. She's one of the most fierce daily leaders I know. She's had previous roles at Moo, Dojo, and Hilton, hotels, building data teams and driving value with them. And today, she's gonna talk us through how she's done that many times over, how to drive data team value when you're first into your new role. Lina, thank you so much for being here. Welcome to the show. Thank you for having me. How's that for intro? Yeah. I I would say you compete with me on the badassery, but I'm not on that one. I'm not sure. I I, I'll put it this way. I think you, I love working with you because you absolutely drop have no BS at all in what you do. Everything you do is is really, really great. So the first thing I want to ask, yes, it's just a nice breaker question, which is already fun to get different answers. Just tell us the first BI tool you started using, maybe loving, that isn't Excel. Everyone says Excel. Yes. For me, it was it was Google Analytics. Like, OG version one Google Analytics, and it was the love of my life. I did every course. I showed up to every meeting with it, taught on it. Like yeah. That's that's a good answer. I love that answer. That that suggests that the marketing, Ben, I know that you have a bit of passion for. I'm gonna ask you what you think about GA four. Let's park that one from the, like, the outtakes perhaps, but, I love it. I think that's a good one. It's amazing how much you can do with that tool. If people if people are data analysts and they're sort of stuck in SQL, Google Analytics is a tool which is it's so powerful. It's an enormously powerful tool. There's a reason Google will do what they do in many ways. It's another example of that. And and so many of, like, analysts discount tools like Google Analytics because, you know, you're not doing raw SQL, but you are doing analysis. It is still data science. It is still drilling into, like, the five whys. And, you know, my one piece of advice for for any analyst starting out would be, like, pick up one of these tools because they are the foundation of what so many of your stakeholders actually deal with because they're not seeing the SQL. They're not seeing the Python that you write. And and it's, I must suppose I haven't looked at the GA four UI since they launched it. I'm I'm in SQL. But, certainly, the g like, it's also a UI which has been battle tested. Like, at every single possible use case, they have thought through and built out with custom UI because they know the answer. They can't give you SQL, so they have to get so it's amazing to see how battle one is because they can answer all the key questions any organization can generally want. It's very painful to get there, but it's all in the UI. It's an amazing breadth of UI there as a as a bit of a Chatter online against GA four is, is pretty bad. But, you know, that doesn't mean bad tool. It just means people have a lot of strong opinions. We won't Google it. Hopefully, YouTube and they upload this won't won't won't block us for that particular comment, but I got it. Yeah. Yeah. Sure. Well, there you go. That's a good start. So let's move on. I wanna set the scene. Currently, you're at Bumble. Obviously, you've run like, formed and run data teams at different organizations as well. We'll dive into that later. But just give us a sense of Bumble now. Like, running analytics there. Tell us the data stack. Tell us the organization structure. Like, how's the kind of latest scene for us about where you are at the moment? Yeah. So, just just to start us off, so Bumble is the dating app. It's also used as, a friend's friendship app, Bumble BFF, and, there's also a business component to it. We also have Badoo, which is another dating app, and then a couple of other apps, including Geneva, which is more for, like, community building. So as you can imagine, we deal with vast amounts of data of people, you know, swiping on each other, chatting to each other, doing all sorts of stuff in in and out. So, we have a we have a big data team. Our our data team now is composed of, analytics and data science, which is different types of analysts in terms of, skill sets, but focused on primarily, like, the analytical side. There's some core data science skills that are, everything from just, like, the ML side of training and deploying, models, but also, like, just deep advanced statistical analysis of of different trends. And then we have analytics engineering, and then data platform. The way we're structured is we have, analytics and data science and analytics engineering together under the operational function, and then data platform and machine learning sits, with engineering. And it's great actually because the teams work really closely together, but the operational alignment allows us to be really close to the business and feed that back to the data platform, which can tend to be somewhat removed. But we run, you know, very closely as a as a team. In terms of our stack, so, when I joined about two years ago, we were running on some pretty out of date infrastructure. We were, fundamentally on prem, and, we have been on a journey for the past two years to level up, and bring on new tooling that's, more relevant for the analyst talent that we have and for the needs of the business. So, we're now in a position where we have, Looker as well as Count. We use DBT. For all of our analytics engineering needs, we use Monte Carlo. And there's quite a few more tools that we're currently exploring to just continue the modernization and to continue to increase self-service, but also just enable productivity within, the analysts and data scientists. That's a great summary. Yeah. It's amazing, actually. Like, though, what what's amazing about that story is, obviously, you're now on, like, top top quality tooling, like, count, place to drop. But, actually, like, the Bumble's done a huge amount with relatively old tooling. Like, it actually in fact, you've it's not like the best in class tooling was necessarily the biggest differentiator of performance. It maybe it's obviously got a lot better since then. I'm sure it's drawing even more value, but I guess one of the things I noticed when I first started talking to you actually that the Bumble had grown a lot with a relatively old stack and done very well with it to up to a certain point, which is kind of interesting thought, I I always thought, in some ways. Yeah. But with all this that's a Exactly. And so, like, there's all this new tooling. There's also a culture component to it. Right? So it's it's what's the old saying? Like, you can change all the tech you want, but until you change the people, it doesn't really matter. And with with changing towards this, like, data heavy focus, diagnostic focus, and enabling self-service and, you know, enabling, like, just quickness of insights and being able to act means the whole business has had to really lean into that. And that, I think, is what has driven the pace and the momentum of being able to change so quickly. Yeah. That's really interesting. It's also in so interesting. I'd love to ask a question about it if I if I can to help people. Like, you mentioned you have, like, sort of very advanced machine learning deployment of models and analytics kinda separate. It's kind of kinda normal to have that. Is that have you really got a clear defining line? I know that everyone strays into both buckets. Right? Analytics can stray. Data science can stray. Is there is that is it quite a defining wall, or is it actually quite fluid? I would say it's pretty fluid. I would say it's pretty fluid. I think that just like with any company, right, like, the opportunities are there if you take them. Right? So if you are working on something that is a deep dive, but the outcome of it is, do you know what? This could actually be, like, operationalized as a model and, like, we could push it into this and this and this domain. I have the expectation of my analysts to be able to identify those opportunities, and if they are so inclined to advance in that way, to take that. Right? And if they're not, because they don't have the interest, don't have the skills, or whatever, then either put it on your growth path or, like, feed it back to the team as an opportunity that has been identified that we need to do something about because it will bring business value. But that's kind of how we operate the team. It's like, you know, the the analysts are out there to to to help support the strategic partners, so all these stakeholders, but also to identify these opportunities. Yeah. Yeah. That you gotta have that full end that full from deployment model all the way through to the stakeholder and the and, like, the business problem all the way through. That's cool. That's cool. Thank you. That's certainly seemed really well. I mean, so what we're gonna talk about today is, links to all the things you discussed, Eddie. You discussed, like tool stack changing, organizational change, organizational layout. All these things are part of, like, how I guess, you as you joined Bumble as a as a data leader and started running those things, it's a kind of big it's been a big lift. And I just wanna you spent some time with you talking through if a new data leader is going to join an organization, how to approach what can feel like a pretty daunting task. All those things by themselves are big. Doing them as a new role in a new organization at once is amazing. So let's maybe spend some time talking about that. And just to bring up so just to remind people as we talk about this podcast, we're we're focusing on data team value, and our framework for that is the tenants, the tenants of data team value. So there's four tenants. There's operational clarity, the idea that data teams should be driving signal from noise, making the business feel simple, not just producing loads and loads of reports, which mean nothing, solving business problems, not just being a data portal, spending time with the business, minimizing decision time rather than just being, a kind of a data product environment, and then measuring yourself rather than doing things the data team can't see. You're instead, showing value and driving ROI. And, really, that that fourth one is really what we're talking about today, particularly under the lens of joining a new organization. So, I mean, Lina, I'm not thrilled. You you've you've in many different industries and many different times in your career, you've come into an organization and led teams and driven organizational change. Like, as I said as we've discussed it, it's one of those things which is, it can feel incredibly daunting to join a new company that way with that kind of leadership responsibility, and not least because measuring impact, proving what you're doing is drawing value is always just an age old question with data teams. How do I prove ROI? And so, I mean, it just makes that that kind of expectation stuff at the beginning is just really difficult. How how do we even think about it? How have you done that in Bumble and maybe in other roles as well? How have you, like, led that change? And we can just drill into that for a bit. Well, that's that's a loaded question. Right? So let's let's maybe start from the beginning. So I think let's just have a situation. You are a new leader. You have joined a company, and all you know is what's on the job description. Right? Like, say it's, you know, senior manager of data or head of analytics, whatever it is. Yeah. You don't really know, like, what your aim and what your outcome looks like at this point. You know, you may have had some, like, onboarding from your manager or whatever, but, like, you don't really know which way to move. Like, do I need to handle the data stack? Do I need to look at the people? Are there specific projects that I'm I'm going to be, like, deployed to? And it can be really overwhelming at the beginning to kind of identify what is the number one thing that I'm going to do that plays up to my skills but helps drive the most business value. And what I would recommend for anybody starting out is take a really deep look at what you're working with. And what I mean by that is at the you know, your first month, everyone just throws everything at you. Oh, you're new. You're like a shiny new toy. Like, you could be on this project. You can help here. Oh, you're a data person. You can help with that. Take everything. Yeah? I mean, it's also really important. Right? Doing that kind of meet and greet, but just being completely blinded by what everyone's saying is part of the task, I guess. Spend your have a ninety day plan, and your ninety plan day plan is this. First of all, focus on your stakeholders. Meet absolutely everyone. I'm talking do not do work. Like, go and meet as many people in the business as possible. Head of FP and A to the salesperson. Go shadow an ops person. Go sit with the product team. Get to know the business. Understand how people view data, how they use it, and what their biggest gaps are, and note everything. Right? Like, take vast amounts of fill up a whole notebook of your insights from the conversations that you're having. Basically, conduct interviews because when you're new, you have the direct, like, ability to ask those kind of candid questions, and you can play it off as, well, I'm new, so I'm just asking this question. Use that to your advantage. So do that. Spend a lot of time with stakeholders. Have this anecdotal stuff written down. Number two, spend time with your team, and don't just get to know them personally, which is a given, but get to know the projects that they enjoy, the projects that they thought they brought the most value in, the projects that, they think they grew the most by in by doing, the projects they hated. Have a very well rounded view of the team, but then through that, be able to analyze what their actual competencies are. This analyst, you know what? They're a great communicator. Technically, they're a little bit less so. But, you know, you start defining, like, who is good at what because that starts to formulate a view in your eyes of how can I combine different skill sets to be able to push a project along? So if I if I have a project which requires a lot of cross functional collaboration, but also communication and socialization with, I don't know, execs all the way down to every layer of the business, I might want to pick an analyst who or two analysts. One who's like the technical counterpart who will, you know, help, do the analysis, do the modeling, whatever, and then a more, like, commute communication heavy expert who can help translate that to the wider business. So you need to do that assessment of the competencies of of your, team as kind of your second step. Then your third step and this is by by the way, you have to do this in your first ninety days. You if you take any longer than that, you will you will not do it. Right? So you have to grace window. You've got a grace window where you can play the naive card and just learn and listen absorb, and then people are gonna start expecting things. Yeah. Got you. Yeah. Ninety days, you're OG, man. Like, you're part of the you're part of the you're part of the building. You you can't get away with it. So the third step, though, is is the pivotal step, and that is the survey. Right? And, the survey is intended to be sent to the wider business, and it needs to be brutally honest, and you need to be prepared to receive the kind of feedback you're going to receive. For some businesses, this might be the first time that kind of survey has been conducted. I know in my previous three businesses, the first time I set the survey out, it was brutal feedback, and may or may not have made me and my heads cry. And I may have considered switching careers, but you know what? It sure as hell set the baseline of what things I need to do in order to progress the team. So let me let me tell you a little bit more about the survey. Please. Yeah. The survey. By the way, just to clarify for everyone who's listening, Lina has given us a template of this survey, so you can go do this yourself. We have a complete playbook on this survey that you can find in the show notes. So with that in mind, now go through the survey, the the gold survey. So the survey needs to focus on what you think are the analytical maturities of the business and where you think you fit into that. So maybe I can take a step back and describe what I mean. Because you've done all these interviews with state or analysts, you're getting a clear sense of where the business is at different domain levels in terms of their analytical maturity. Maybe this is a good time to show, a view of what I mean by analytical maturity. Yeah. Yeah. Let's bring it up on the spring up the slide. That's wonderful. So Here we go. There you go. This is the analytics maturity curve. And, basically, what it does is it it describes the movement of data through an organization in terms of its complexity of usage. So, you know, at the beginning, you just have the data. So if it's sitting in servers, it's sitting in warehouses, wherever, then you do something with this. You transform it. You cleanse it. Maybe you start automating stuff. Then you actually surface it for people, you know, whether it be Power BI, Tableau, Looker, whatever account, but it's it's being shown to others outside of the data team. And then you keep progressing by adding more and more. You you you start, like, actually using the data towards other things. So maybe it's a statistical model. Maybe it's a deep dive analysis. Maybe it's, something that self replicates, maybe it's a machine learning model, maybe it's, etcetera, etcetera. And as you move up the maturity decision output. You've got, like, raw data through to what's going on to decision, some degree of automated decision or manual decision making. Yeah. Got it. This is awesome. Yeah. And the the last part is data is actually linked to ROI, meaning you've created data product and it is self generating of some sort of business outcome, which is could be a cost efficiency or could be a revenue growth. So something like a recommendation engine is monetizable because it is a data product that enables you to grow revenue by x amount. Right? So Yeah. You have about this curve. And when you're doing all these interviews and and and these assessments, you have to take a deep look at where you think the company is and different domains, product marketing, what revenue might be in different areas, but have, like, a little table that just gives your, perspective on where you are. Yeah. And there are assessments online that, like, we can link to where you can do this in a more, like, know, formulaic approach. But your anecdotal evidence will will suffice for your take. Right? So Yeah. Like, the rule of five kinda thing. Like, if you see a pattern of four out of five times, it's gonna be pretty statistically likely to be a real thing. Exactly. Exactly. And and then when you do the survey, you cater it towards where you think you are in the maturity group. So, you know, you send out a survey. Let's say you think you're more kind of in the self-service and maybe you've got some stats going on in some places. Then you might ask about, you know, how impactful do you find the data team? How much are you using the data to, do your day to day job? How many times a week are you blocked by the lack of data? What do you think about the services of the data team? But you need to get kind of your data team on board with this and, like, talk about why this is gonna be valuable. And what I would recommend and and I've definitely instilled in in my teams is the baseline outcome of these surveys, which you should do on a regular basis, is what is what composes your, OKRs or your objectives or whatever your company uses. So you should measure yourselves based on the outcomes of the survey. Yeah. See, this this survey basically becomes like a baseline of your business. Like, the first ninety days, we know we've got a sense of where the biggest areas are. We've baselined ourselves, and you can show improvement every quarter, whatever how often you do it. Is that basically the idea? I guess just to help, like, the I I guess you want us to be anonymous, but I'm assuming you wanna get a sense of, like, which part of the organization is answering the certain way because that helps you understand, like, how to engage with the right bits of the business, the right level of maturity. Because not everyone's gonna be straight to predictive models. Some just need access Yeah. In a different way. So it just gives you that whole map of, like, decision types workflow problems that are coming up or at least gives you a big heat map for that, which is amazing. That's awesome. Exactly. And you should like, you know, if you're using Workday or something similar, like, should be able to integrate, like, what department the person answering is on. I would advise making it, anonymous because you want the most brutal feedback Yeah. That you can get, and you also have to be prepared for what you receive because it might not be one of the most interesting things is, like, I think a lot of data teams have a self perception that is different than the business perception. And Yeah. Yeah. Really helps dispel that. And I love it. It's not you're not letting the business dictate what your value is, but this is just an input for you to work out a strategy to move them on in the way they might not even know they need to be. What Exactly. Generally, I find data teams forget is that they get stuck on one of these areas, like self-service. I must enable self-service, and they forget there's a whole distribution of workflows, whole distribution of decision types that they could be optimizing at different levels of maturity, different levels of detail, strategic to operational. And this survey gives you that full heat map, basically, of a way to see all the different types of data engagements and where the problems are that you can drive impact to. This is this is really cool. This is really good. And so that you've given us I think, actually, previously, you've given us, a bit of a deep dive and a template that people can use to build their own surveys from, which you can find in the show notes. It is an amazing document. I really can't recommend enough that you take that up. Even if you're not I guess you're saying even if you're not new in the first ninety days, you could still adopt this kind of strategy to start baselining for yourself where you are. Right? It's not a just a ninety day thing. As good as that is to do then. And especially if you're if you're doing a lot of modernization work where the impacts aren't gonna be immediately felt by the business, It's a good it's a it's a good way to measure, like, the impact that the business perceives out of the work that you're doing. I would also say that the survey and the interviews are alone not enough to measure true data impact. And and there's a lot of literature on this, and everyone has a different opinion. But there is something to be said about really getting your analysts and data scientists, whoever it is, to get that feedback loop from stakeholders on stuff that they produce, whether it's a data product, dashboard analysis, whatever, and understanding the impact it made on their decision. So say you've put together, I don't know, a geographical dataset that is now being used by marketing for hyperlocal, like, marketing campaigns or something. You wanna know what that drove so you can go, hey. We built this dataset. It enabled marketing to drive x amount of revenue as a result. Yeah. And you can use that in combination with the outcomes of the survey to be like, here's the revenue I'm driving, here's the cost efficiencies I'm driving, and here's what people think of and use our data for. I mean, that's that's great. I I opt the ROI question is a is a never ending question, and I don't think it should ever go away. I think well, I what I've observed myself is that, the people who are trying to attempt some degree of measurement, be that quantitative, in fact, it's often better qualitative in some ways than quantitative, are the ones who who don't need to justify the ROI as much as the business. It's the ones who don't do anything or try and get overly scientific where you just get lost in the weeds rather than just actually just being open to feedback and just thinking about it. Like, it's a continual it's a kind of it's a it's a principle rather than a number in many ways, and I think that's Yeah. Placed that really well. That is that is so great. Thank you so much. The last question I wanna ask on this topic before we move on, if I can, is just expectation management. That's one of the biggest worries I think people have when they come into a new role. I often hear is people saying, how do I manage the my leadership's expectations of me and how fast we can change? Have you got anything on that which is, like, particularly important? Maybe it's like that first ninety days. How do you make sure that your exec team or the people that you're reporting into about data recognize and baseline their own expectations because, you know, they may be good looking to go very quickly, but maybe that's not realistic. Have we done that before that's come up? I think the the the key thing and the thing I think many analysts and analytical people struggle with is over communication. And I giggle because it's the one thing that my own leadership team, they always laugh because I always say, how many times did we communicate about this? How widely did we socialize? What's the frequency on any topic? It's it's the thing that I think people die by. Like, if you if if your execs don't know that something's delayed, don't know the timelines, don't have a deep understanding. They're never gonna have a deep understanding of, like, the day to day of what it takes to get something across. But if you can have a timeline and set expectations and communicate when those expectations are not gonna be met or there are delays and you can be very clear around the why, they won't mind the delays. Like, I think people are very afraid of actually bringing forward mistakes or bringing forward, communicating around delays and just hiding it instead. But, like, actually, execs appreciate over communication. They're so busy in their day to day. They're gonna read probably the first two lines of whatever message you sent them. So, you know, declare and communicating that, like, hey. This is what this is about. Read it in your spare time. But I I can't, like, nail this home enough. Just the visibility to XACs is what they value more than anything else. Love it. Thank you. That's great. I when we while we're on, we've got we have time left. I'd love to just ask you, thinking right back to the beginning of your career, one of the things that people might wanna know about you is that I think you started off as doing medicine and then switched to data, which to my wife, who's a doctor, is very intriguing to hear about. So maybe you can like, when you first started out in data, what's the piece of advice that you give yourself that you now know about how to succeed in data? Maybe you give other people that you you're mentoring now. Well, funny enough, I used to say that the reason why I switched from medicine to data is because I don't like people. I like numbers. But as I grew in seniority, I realized that, actually, you really have to like people, in order to do numbers. So, I can't use that excuse anymore. But anyway You can just speak to people for the first ninety days of a job, so you literally can't backtrack on it yet. Yeah. Exactly. Exactly. Look, I've never looked back. I I think this has always been my career path. I just didn't know it. But anyway, one of the things that I did when I was starting out was I spent a lot of time in an agency setting, and I I was exposed there to multiple industries, datasets that were completely, like, disparate from each other, tools. And I had to navigate kind of every aspect of data from data end to AE to, you know, more complexity of science, etcetera. And and I actually think that that time spent there with the exposure of and and the variability, so you you might get that in in consulting, you might get that in an agency setting, incubator type of stuff, is really valuable because it just it trains your mind to be able to, apply core content concepts to, the similar systems, if that makes sense. So, like, this is why I laughed at my first tool that I I knew was, Google Analytics, but the reality was it was also Amplitude and Mix panel and, you know, Adobe Analytics and etcetera, etcetera. And you kind of approach analysis and visualization and what whatnot across those tools in a similar way. But if you learn one too well, you can apply the concepts across them all. So I would encourage all analysts starting out, like, just being exposed to us. Don't get yourself pigeonholed into, I am just this expert in this in this area, unless you very early on have decided that you wanna very much specialize in a very niche field. That is an important differentiator for for people. Like, do you want to be a more generalist analyst, or do you want to be a hyper specialist who does, like, you know, Adobe DTM tracking tags? Yeah. That makes total sense. Thank you. Right. I've got one more thing for you before we go, and that's our anonymous data letters from the community. I've asked people that told me we were that we were interviewing you, and I got a few questions back. And I love this one if I can ask it to you. Are you ready? It's, here's the paper. Here's the question. I'm an analytics leader running a small team of analysts. I've become increasingly convinced over the next three to four years, my EQ, I guess, emotional intelligence and soft skills, are going to be more important for me and my team than technical ones. Chat, g p t, anyone. What are the most important soft skills Lina thinks my team or any team should be developing? I guess so how does that sound? It's a question. Soft skills. Yeah. I mean, number one is what I said before, which is being a very strong communicator. That is the number one skills. And a communicator that's able to flex based on seniority levels. You speak differently to an exec than you do to a VP or director than you do to, you know, an IC, and you should recognize what it takes, to flex between those levels. Number one is communication. And I would say number two is, like, being an opportunity seeker and an opportunity solver. So a lot of people fall into the mistake of, I've identified this problem. What do I do about it? Don't be a problem identifier. You're an opportunity seeker and a problem solver. Love it. If I could throw my own two cents on this, I would say the thing I see most differentiated for people in soft skills is just problem critical thinking and problem solving structures, like just structuring an argument, structuring methodically. It just to like, commute part that's communication. Communication to yourself and communication to others about how why you've got an answer. That's great. Thank you so much, Lena. It's been a such a joy to speak to you. Thank you so much for being here. If you wanna hear more about from Lena, you can find more about who she is, where she is on LinkedIn, etcetera, in our show notes along with Lena's amazing stakeholder survey. That is a thing you must try out. You can follow about that in the show notes. Check us out at cal dot co slash m t n, and, also, feel free to check out Cal. Thank you, Lina. Thank you. Thanks for having me. Me.