Naomi Smulders, a Data Scientist at Online Dialogue, joins us to discuss her path into data science from psychology and cognitive neuroscience, how a process mining approach can be used to analyze the customer journey of web users, and the importance of building and growing communities.
Naomi is a Data Scientist at Online Dialogue, a small online experimentation consultancy firm, where she designs new data science products in an interdisciplinary team of psychologists, designers, and developers. She has a background in Psychology (BSc), Cognitive Neuroscience (RMSc) and an Engineering Doctorate (EngD) in Data Science.
Naomi likes analyzing large quantities of data to gain insights into human behavior. Alongside modeling and storytelling, Naomi enjoys teaching and giving presentations. She has previously taught statistics courses and training and has presented at several national and international conferences and meetups.
Naomi is a member of the Outlier Events committee and has helped to organize the first two editions of the Data Visualization Society's Outlier Conference. She also co-founded DSconnect, the alumni association for the EngD in Data Science program at the Jheronimus Academy of Data Science.
Tags: data science, process mining, customer journey, web analytics
[00:00:08] Lauren Burke:
Welcome to Women in Analytics After Hours, the podcast where we hang out and learn with the WIA community. Each episode we sit down with women in the data and analytics space to talk about what they do, how they got there, where they found analytics along the way, and more. I'm your host, Lauren Burke, and I'd like to thank you for joining us.
Today, I am so excited to have Naomi Smulders joining us. Naomi likes analyzing large quantities of data to gain insights into human behavior. She is a Data Scientist at Online Dialogue, a small online experimentation consultancy firm, where she designs new data science products in an interdisciplinary team of psychologists, designers, and developers.
Alongside modeling and storytelling, Naomi enjoys teaching and giving presentations. She has previously taught statistics courses and trainings, and has presented at several national and international conferences and meetups. Naomi has helped to organize the first two editions of the Data Visualization Society's Outlier Conference. She also co-founded DSconnect, the Alumni Association for the Doctorate in Engineering and Data Science program at the Jheronimus Academy of Data Science.
Naomi has a Bachelor's in Psychology, a Master's in Cognitive Neuroscience, and an Engineering Doctorate in Data Science.
So welcome Naomi. Thank you so much for taking the time to join us.
[00:01:39] Naomi Smulders:
Thank you. So glad to be here.
[00:01:42] Lauren Burke:
So to start off, could you give us a little bit more information about your background and the path you've taken over the years?
[00:01:49] Naomi Smulders:
Yeah. Cool. I was born in the Netherlands, and I did bilingual education in high school, which means that I did a lot of English. And then they gave me this opportunity, like, you can go anywhere in the world, so I decided to move to the UK when I was 17 to do a psychology degree there. Because I always wanted to figure out how people work, what makes them tick.
And people are just the most fascinating creatures on earth, I think. So I really wanted to study kind of like what motivates people, how do they behave, what makes them groups, what makes them social, what makes them anti-social, mostly. So I did my undergrad in the UK. I started as a forensic psychologist, but it has a lot of law in there, so I kind of switched to just regular psychology.
And then I found cognitive psychology and neuroscience along the way. So after my bachelor's degree, I did a master's degree in cognitive neuroscience. And I mostly chose this path because psychology is well and good, but there's a lot of things in there that are just theory and really hard to test. And that always kind of like nagged me, like having just this kind of like global theory and not really having the data to support that theory.
So I tried to find the most kind of like data-driven part of psychology that I could find, which is neurophysiology or cognitive neuroscience because it's about, I don't know, blood measures and EEG signals and MRI signals. So it's like measurable just gives you a data point, and something goes up or down.
So when I did my master's degree, that was really cool because now I could actually measure everything that was happening in the brain instead of just kind of like having this global theory that had so many variables that you couldn't even comprehend. So I really enjoyed my master's. I did a research master's, part of which I also did at Cardiff University in the brain lab there, in which I did some neurofeedback studies. So with MRI machines. So I really enjoyed the kind of like data-driven part of like figuring out the brain.
So to continue that, I went on to be a researcher at the Dutch Cancer Institute. I did that for like half a year and we tried to get a grant to apply machine learning algorithms to predict cognitive decline in patients with chemotherapy. Because the reasoning was that some breast cancer patients get chemotherapy and then afterwards they get cognitive problems like concentration or planning or, yeah, just all those kind of. Memory is often cited, but not everyone gets those. So we were trying to figure out if we could find a model to kind of like predict who would be having these problems and who wouldn't. Which I thought was a super cool thing to research.
But getting the funding around was really hard because we wanted to apply kind of like modern techniques, like machine learning and advanced statistics. And the funding body was like, no, it's too, it's too out there, too innovative. Like, you could do this with traditional methods. So yeah, you have to change that. And then I was like, but this is what is really the cool part, like the, the statistics and the methodology and stuff.
So that's how I ended up at Eindhoven University, to do a engineering doctorate in data science. Because then I got to learn all the cool stuff and the methodologies, and the statistics. And especially because for my masters I had to do a lot of statistics anyway, which I enjoyed. Which made me one of the few people in my master's program who actually liked our statistics classes.
So yeah, for my interest in statistics and always wanted to really be able to measure the things that were happening, I thought data science would be the best, like kind of like application of that because it just gives you the tools to measure all the things that you would like in the best way, like that we are currently capable of. So yeah, basically kind of like went from like, trying to understand that, trying to find the best way to measure and yeah, going from there.
So now I've been working for a Data Scientist as for three years. And currently, I'm at Online Dialogue, and we're a small marketing consultancy firm. Mostly figuring out how to optimize workshops and behavior.
So we do a lot of web analytics but we also have some psychologists working for us. So I basically found a job in which I get to combine my love of like trying to understand how people work with, like, doing it in a data-driven way as a data scientist. So yeah, I found a job that kind of like incorporates all my previous studies into one role, which is really awesome.
[00:06:24] Lauren Burke:
You really have a really really interesting background with just all the paths you've taken. But it sounds like one of the kind of key factors is that you really like working with data in a context that allows you to better understand human behavior. So how does your background in psychology and cognitive neuroscience affect the way you approach a data analysis?
[00:06:45] Naomi Smulders:
I think in multiple ways. For one, it's like what you really learn in psychology is that things don't always are what they seem. Because especially in human behavior, people say they do one thing, and then they actually turn out to be doing something else. And I think it's the same if you are approaching a data project because the first figures that you get, you see something and then you try to figure out like, I am making an assumption what I see, is it really what's there or are there kind of like some factors lying underneath that actually can explain this result or contribute to this result? Or is it really the full picture that I'm looking at? So that's the one side kind of like to go into the next level be critical.
And I think the other practice that I really, really take, especially my consultancy about data is what I know about human perception. Because if you have figured out everything and you know the whole story of the data has to tell you, and then you need to tell someone else. Because they probably need to do something with this knowledge.
And how people perceive information, take in information, how much they can remember at the same time, what things like makes them kind of like remember things or forget. It really helps in the storytelling of like what you're trying to show with the data that you researched or the problem that you tried to solve and you're bringing the data for.
So yeah, it's a bit being critical and also a bit telling the story well so that everyone understands.
[00:08:08] Lauren Burke:
That makes sense with the storytelling aspect because you kind of understand what people want to see and what will make them potentially feel more open to accepting the results of this analysis. One thing that we see a lot with data and modeling, right, is there's a ton of bias, and we can introduce bias without even realizing it.
Do you think your background, understanding some of those cognitive biases, helps you to better avoid some of the assumptions you might make on top of those?
[00:08:35] Naomi Smulders:
I'm not sure if I'm better at avoiding them, but I might be more aware that they're there. So I mean, some biases are just so naturally ingrained that it's just, it's really hard to avoid them. But it's good to be aware of them so you can research like how big the problem is. And I think, yeah, that's definitely one of the things.
But also I learned to appreciate because in the end, bias is also some kind of shortcut. So it also kind of taught me to appreciate that sometimes the easy methods are also very good. Like you don't always have to have a very complex model to make a pretty good, accurate prediction as your brain does. Because if you, what a bias is is nothing else then as kind of like a simplified model of, of a complex structure so that you can make quick and easy decisions.
So your brain has this thing down ridiculously well because it's very, very efficient and making quick decisions all the time. And then it gets it wrong, but the margin of error is okay. And I think also in analytics we can apply this a lot by trying to find the most simple model that we can apply and not having to complicate it a lot further without gaining actually extra like predictability or accuracy.
[00:09:47] Lauren Burke:
Right. Like Occam's razor. That's a principle I feel like the longer you're in a analytical or data science role, you really start to follow that and abide by that because like a lot of times you don't need to build a neural network for
[00:09:59] Naomi Smulders:
[00:09:59] Lauren Burke:
Every single thing you're looking at, right? A linear regression, a logistic regression is oftentimes good enough, and right it's such a developed, well-developed approach that you kind of understand that it's going to be a better option the majority of the time.
[00:10:16] Naomi Smulders:
Yeah, and I think this is the beauty in how both analytics and cognitive neuroscience kind influence each other. Because we build neural networks because that's how the brain works. And then we learned how some of our neurons in the brain actually work because we simulated it in neural networks.
So it's kind of like the symbiosis of these two fields. Which is super cool that you can learn from one another and like optimize the knowledge in like both fields by kind of like looking across the border in someone else's field. This is really cool.
[00:10:44] Lauren Burke:
Yeah, that makes sense. You definitely understand really well how neural networks are connected to not only data science, but basically the way they work in a human brain and how they're replicated over.
Are there any other skills you develop during your studies that you've intentionally brought with you into your data science practice?
[00:11:03] Naomi Smulders:
Yeah, mostly it's just my statistical thinking and how you approach a research project. Because especially in my bachelor's and in my master's research program, for me it's second nature that you don't just take everything for granted and then you put in hypothesis and then you put up an experiment and put something to the test. And you try to find the flaws in your reasoning by looking at it from different sides and talking to other people to see if they can find the flaws in your reasoning.
So yeah, to actually build a solid case. So gather all the evidence, find all the data, and also build the story, but also kind of like get the case and be open to critical notes from either yourself by taking the devil's advocates approach or other people by showing people your work and being like, "Okay, try to find the flaws." So that you can build a stronger case. And I think this is definitely something learned in my studies because there's a lot of group work and yeah, the whole scientific field is all about challenging views.
People say something, you write a paper about it, and then there's four other people trying either to replicate what you found to make sure it's true or finding something different that kind of questions what you found. So I think this whole kind of like back and forth and this constant critical acclaim is something I really do in my work and also like a lot. And I think sometimes the business could do more of that.
[00:12:29] Lauren Burke:
Right. It sounds like you've brought kind of two of the hallmarks of academia, the scientific method, and then the peer review process and embedded that in your own data science work, which I think is great. I think there is a lot of transition that could happen from academia to industry. Oftentimes it's so separated that when you have someone that's been in both of those areas, it's easier to see what you can bring from either side to make both better.
[00:12:55] Naomi Smulders:
Yeah, exactly. And also especially because like the scientific method is so ingrained if you did the scientific course, but it's something that can be so new to people that come from different backgrounds. And I always really like also teaching that. So just telling people about that and making them, like, see the reasoning. And people usually get very excited if they can like, go through this and be like, "Oh, I put an experiment up, and then I got this result, and now I can like apply statistics. And now it's true."
And yeah, it's nice because it's a method that gets a lot of people excited, but people can be a bit scared of it at first because it's science. It sounds scary. It might be like data and numbers. But actually, a lot of people, once they get into it, they get really excited about it, which is always what I really enjoy seeing.
[00:13:41] Lauren Burke:
Right. It's a good framework to go by. And the more you understand why each step is important and how it can help you, I feel like you're more and more open to that becoming kind of your standard approach, which is great.
[00:13:53] Naomi Smulders:
[00:13:53] Lauren Burke:
Speaking of instruction, you've taught some courses and you've shared before about how companies can use a process mining approach to analyze the customer journey of their web users. Could you tell us a little bit more about what process mining is?
[00:14:08] Naomi Smulders:
Yeah, sure. So we did this project as a new project at Online Dialogue, where I'm working right now. Because process mining is actually a data science technique, and it was first introduced to find bottlenecks in a process, or kind of like usually a linear process.
So for example, if you're applying for a loan at the bank, you fill in the form, someone has to review it, it either gets like accredited or not, and it gets sent to the person that's actually lending out the money. And then there's this linear process like that has multiple steps. And then there are some criteria. And process mining was a way to kind of like find where the bottlenecks are. So which step takes a long time? Where are the delays? Is there some feedback loop in there that just can go on endlessly so that you never get anywhere, which is a high frustration point for one of your customers. Because if someone says like, "Okay, you go there and then you go there" And the person points you back to the original source, then you get this loop.
So in this way, the process mining technique tries to kind of like take this journey. So all the steps that you have in a customer journey and figures out how long do the steps take, which is the order that the steps are taken into and so on. So this is a fairly old, well, fairly old, I think it is, been around for like 20 years or something.
And I try to apply this to web data. Because in web analytics, what we see on a website is that people also go from, for example, you get to the homepage, you go to product page or listing page where you click on the product that you'd like, you add it to your card, and then you go on to check out. Well, this is a fairly linear process.
But I don't know if you have at some time, done some online shopping. You never just go to the homepage, pick one product and then buy it. You probably like go to the homepage, first look at the deals because you know maybe there is something off in the discount or on sale, and then you go back and then you try to find maybe that like dress or something you were looking for. Then you click on all the listers. Then you kind of like filter out some of the things that you do want. Maybe you want long sleeve or short sleeve or whatever. So you get all these different kind of like options and you go see a couple of products and then you go back.
So if you do this, you'll probably be doing something different than if I was doing this because I'm looking for a red dress, or I might be buying shoes or whatever. So for every user that's on your website, you get this very unique journey of people going through all the different pages of your website. And only at the end do you actually see the pages that are mandatory to pay is where everyone kind of like builds up.
And what we try to do in your analyzing the customer journey is try to find where do people find the information that they need? And also what are the bottlenecks? So what makes them frustrated, and makes them leave or makes them do something else. Or what takes a very long time, which should be going quickly?
And in original methods, you just make a funnel. So you say, "Okay, ten people go from this product page, and then five people put it in the cart, and then three people pay for it." But then you don't know where the other people went. So if five people move from one page to the other page, where did the other five go? If you started with 10 people?
So what process mining does, it kind of like collects all these different journeys of all these different people that are doing their own thing on their own website according to their own needs and tries to make sense of all these pieces of information. And this is especially something in web analytics that is very hard to see because it's so diverse. Because you have, like, if you have 600,000 people, you have 600,000 unique customer journeys. And it kind of like tries to find the most common steps in the paths of these people.
So it could be that, for example, you see a lot of people going to the search bar to look up a particular thing, and then having to click two or three times and then going back to the search bar because they haven't found what they were looking for. But this is just one part of their whole journey.
So what process mining does is actually, it takes all these journeys and overlaps them and looks for kind of like the commonality. So the common steps, the common sequence of steps that's in there. To figure out what people are actually doing when they're not doing the thing that you expect them to do, which is most chronological logical order that you designed your website by.
But yeah, like I said before, people always tend to do something different than you expect. So it's a way to kind of like capture that, like the roaming around, that orientation behavior and see what they're doing there.
[00:18:20] Lauren Burke:
So when you first started looking into this approach, was there anything really surprising that popped out about the customer journey?
[00:18:28] Naomi Smulders:
So we applied this for a couple of our clients. And every one of our clients had different insights, but mostly it's that we found that people had a very high need to have additional information or help, especially when we did this during Corona times. Because now you have to get all your information from the website. So usually, like sometimes you have a webshop, but then there's also physical stores.
Or for example, if you're buying like a quite expensive product or something that you use a lot, like furniture or something. You, if you don't know, you just go to the store and ask the person there, because you expect the person there to know more than you. So you look for advice. But if the stores are closed or far away, you don't do that. So that means that you have to fulfill that need of advice and service on your website.
And we saw a lot of points in our customer journey that the people did some searching. So they looked at products and they looked at product descriptions and stuff. And then just went to look either where the address of the physical store or tried to, I don't know, chat to the customer service. Or found this kind of like explanation and service pages. So from there we get like there's a high need of information. So let's see if we can build some experiments to, for example, get some very clear links from the product pages to this information. And that worked very well actually. So we got a lot more click-through and a lot better conversions out of that because we fulfilled this need. So yeah, we had those.
And we also actually found sometimes just some bugs in the website. So for example, if there was a combination of product, it would just empty a cart without explanation. Which is very frustrating, if you just filled in everything and you're like, "Okay, let's buy it." And then it's like, "Okay, it's empty now." So yeah. Those are to me I think the most surprising.
That actually shows you that where people go actually can tell you some psychological need that people have. But also it's kind of like user-testing your whole websites very thoroughly. If you try to break it there, there's always some people who managed to kind of like do something that shouldn't be possible. So then you go to your IT team and they go like, "But you shouldn't be able to do that."
"Yeah, but we did..."
[00:20:38] Lauren Burke:
That's so interesting. Cause I know a lot of design principles are that you shouldn't have too much information. You should keep it as minimal as needed just to not overwhelm people. But it sounds like you actually found that people actually want as much information as they can possibly get about the product.
[00:20:56] Naomi Smulders:
Yeah, it's a bit of a, because if you overwhelm them, it's just that they need to be able to find it easily. So it's not that they wanna see it all at the same time because then they get overwhelmed and we see that also back again in a lot of our website designs. But pointing them if they need it and, like, making it very easy to get that information.
So either putting the links to the right places and at the right spots rather than putting all that information there. Because not everyone needs it. But if like 10 to 50% of your user do need it, then it's very nice to have that link there. So it's more about accessibility rather than the presence of it on the website itself.
[00:21:35] Lauren Burke:
That makes sense. So just continuing that funnel for those who want to continue on that part of their journey.
[00:21:41] Naomi Smulders:
[00:21:42] Lauren Burke:
That's so interesting. So other than the fact that if you have 60,000 customers, you have 60,000 unique journeys to try and track and aggregate, what are some other kind of challenges you might encounter in process mining?
[00:21:56] Naomi Smulders:
Well, it's a fairly new technique. So basically we work with a new tool that's designed by the University of Eindhoven. So they have the interface. So they make the pretty pictures that shows you, okay, this step, and then that's this page, and then that page. Or this action and that action. And so many people flow from here to there and from there to there. So that's okay.
But yeah, you have to get the data in an event log. Which means that you have to get down and dirty with your data. Because most of it 's in Google Analytics, everything is aggregated. Which means that you can't have one path or one customer journey. So to do this, we had to actually find it. Query our data from Big Query, which is the database from Google Analytics. So it's on a user level. So you don't have all this aggregation that the Google Analytics interface gives you. The average is kind of what is the problem, is that you don't see individual journeys anymore. You just know generic ones. And the whole process mining approach is based on kind of like having all these individual journeys, like as one path and then another path, and then another path.
So I had to brush up my SQL skills. And really get familiar with the GA data. And later also with the GA4 data because obviously Google Analytics 3 has a different data model than Google Analytics 4. Google Analytics 4 is more appropriate for process mining because it's already an event log, so it's easier because you don't have to transform a lot of your data. But yeah, I became very good friends with the Google Cloud on the backend.
[00:23:26] Lauren Burke:
That's so cool. So it sounds like the new development for Google Analytics has been very helpful with this process and has made it a lot easier for you to get that data into a format where it's easy to use for process mining.
[00:23:40] Naomi Smulders:
Yeah exactly. Because especially now that Big Query becomes free with the GA4, Google Analytics 4. So the connection is already there, so now you just need to kind of like, get it out rather than get it out and transform also a lot of things. So yeah, it's been, it's been easier.
[00:23:58] Lauren Burke:
That's great. Are there any other new developments or technologies or tools that you've been excited about for web analytics and data science?
[00:24:08] Naomi Smulders:
Well, that's a very tough question. Because I'm always trying to kind of like find this interface. And I think the most exciting I've been is that it becomes easier to get access to your raw data. Because well we've been using Google Analytics a lot for other tracking data. But what you see now is that actually all the platforms but also all the tracking platforms store other data in data lakes. Which means that it's more accessible on a lower level, which also has a downside because you need to know better what you're doing. You can't just have like already made reports already there.
But for data scientists, like me, it's really exciting because yeah, you actually get to see the raw numbers and get really, really familiar with the intricacies of your data. And as we're moving also to, for example, server side tagging and more and more kind of like owning your own data, I think this is also a very good development. Because yeah, giving everything to Google actually is a lot of work while the interface that we're getting back and the things that you can learn from there are also like reaching the ceiling.
There's a lot of knowledge and insights that we can still gather, but then we have to kind of like, take ownership of that data and really start relying on our own kind of like skill set rather than just Google's engineers and UI builders.
[00:25:25] Lauren Burke:
That's cool. A lot of exciting stuff coming for web analytics and data science.
So, while you were completing your Doctorate of Engineering at the Jheronimus Academy of Data Science, you co-founded the DSconnect Alumni Association. So what led to the formation of that organization?
[00:25:43] Naomi Smulders:
Okay, so when I was at JADS, I was working with people from literally all over the world. So we have a cohort that starts every six months, and that's about 10 people. So it's part of the EngD is kind of like a traineeship, but at a university. So we did multiple projects for multiple companies. So it's a consultancy based but then attached to each project was also a university course. So, for example, if you did data visualization, you had 10 weeks of classes in data visualization and then you did a project for one of the companies that you could apply all those principles in theory to. Which was super cool.
And it's also quite new in the world. And it attracted a lot of people from a lot of different backgrounds. Mostly computer scientists and mathematicians. So I was a bit the old one out as a neuroscientist there. But yeah, I had people from China, Bulgaria, Columbia, Mexico, a lot of Germans, UK, US. So we had this really, really multicultural dynamic and all these people were there for two years and you work with them and a lot of them became also my good friends.
And it's amazing to work with for like two years for at least, with this 30, 40 incredibly intelligent people that have so many diverse skill sets and are so good at what they're doing. And I was like "Okay, this is only two years, but like I can still learn so much from you and there's so much potential here."
So I need to find a way to keep uniting these people. To be able to keep learning from them. Especially because they all went off in different directions, as well. Like not did they only go to different companies, but most of them also went back to different countries. And this feeling of community was amazing in those two years. And I wanted to kind of like build something that kept that community together as good as possible.
Because you can learn so much. And we have shared also so much in those two years because most of them just uprooted their whole life, moved across the ocean and started a new life here in the Netherlands. Which always amazes me, people go here.
So yeah, you have this family and you want to keep them together and share knowledge that we did. Because we used to do this every week in our course and now through the association we can do this by getting together, just telling how we are. We did some hackathons, we had some just events in which we had speakers to come and tell something about their line of work.
And what we do now a lot is alumni talks. Which means that the older generations, so the first cohorts get to tell about their experiences and their stories to the trainees that are currently in the program. Which are like, yeah, four or five years. So it keeps building and you keep spreading that knowledge. And I think that's also for me, a very important part that you share what you learn.
[00:28:32] Lauren Burke:
That's so awesome. I feel like one of the best things about data science is that you have a lot of people from different backgrounds. Even extremely different backgrounds, which it's very interesting to see the perspectives they have and the way they approach a problem. And if you can work on things together with people from different backgrounds, especially data science projects, right? You're getting a lot of different approaches and you're getting a lot of different perspectives combined. So I think that's great that you have developed something that can continue, that you can continue to learn from each other, keep connected, and hopefully work on some really interesting things together.
[00:29:10] Naomi Smulders:
Yeah, it's been a real blast. And also, for example, when I did a project with a computer scientist, like my programming skills weren't super good when I started, when I started this. But in the end you still think the same because you're still trying to solve a data problem. But one person just, I don't know, really likes to kind of like code it up and make some pretty pictures. And the next person likes fine tuning the model and really looking at the parameters and all those kind of things. So what makes it amazing is that they all have these like things that they really, really like, but you still work together on the same project.
So having these kind of like different backgrounds and kind of like looking at a problem from all these different views makes it really, really cool. Because you, you get to very new things that you never thought of if you're just there in a room with all the people that did the same as you did like before. So it's very inspirational.
[00:30:01] Lauren Burke:
Right, and everyone has different strengths they're bringing. So you're getting the opportunity to learn from people and see how people who are potentially very skilled or expert level at some of these tasks are going about it. And then you can take that back into your practice along with all of the different perspectives you've just learned as well. So that's really cool.
[00:30:18] Naomi Smulders:
[00:30:20] Lauren Burke:
So it sounds like community is a very important thing for you. And outside of work you are involved in a number of different initiatives, including as a member of the organizing committee for the Data Visualization Society's Outliers Conference. What inspired you to get more involved in your local and global data communities?
[00:30:37] Naomi Smulders:
Mostly because I just really find people that are passionate about what they do, the most fascinating people on the planet. You always are in a good and happy conversation if you talk to people that are excited about what they're doing because passion is something so awesome and inspiring. And there's a lot of people who really enjoy what they're doing. And people also like to tell about what they're doing, like I'm doing right now. It's a good thing. So, and I wanna channel that.
So first of all, I just get a lot of energy from talking to people who like what they do, because it just makes me happy. So that's the first thing that inspired me. And the second thing, I think, like I said, I think the best ideas come if you combine different disciplines and different views. So I'd like to facilitate that also. And for example, with Outlier, it's with data viz people. And I'm not a great designer, but I like color theory and I like thinking about how to present things to people in the storytelling part.
So I met data journalists there and I never even thought that this was a profession until I was talking to these people and I was like, "Oh, it makes absolute sense that you do these things." And it's really cool because they are the people that get to combine all these different disciplines with design and data and storytelling, and the programming skills.
And everyone has a very different background, but they all make amazingly beautiful stuff and have their own story. Yeah, I really like facilitating these stories being told because I think if you're passionate, you deserve a platform.
[00:32:08] Lauren Burke:
I so resonate with that. I also, I really love data visualization and I am not a very skilled data visualizer myself, because a lot of times what I'm visualizing isn't for anyone else to look at. It's mostly just for me. But I love looking at really cool visualizations. Like I recently learned, there's something called a information designer. And sometimes people that are in that role aren't making something in Tableau. They're making something to print out in a enormous poster size, and that's how you're going to look at the visualization, which is just a really interesting aspect of the visualization field that I didn't know about until recently.
[00:32:46] Naomi Smulders:
Yeah, that's what I liked about Outlier because right now they are printing a magazine, so then all the data visualizations are in print and they're just super cool. But it's different projects from kind of like making a network graph of characters in books. To also data art, in which different kind of like heights of a river was kind of like molded into stone. And people use fields and lights and it's how you can translate numbers into not just kind of like numbers in a bar graph, but like colors and sounds, if you talk about sonification, because it's basically making data into different sounds in which you use like, for example, page frequency and volume to talk about the different attributes of your data. And it's so creative and it's so inspirational to see like there's so much more that you can do to just print the number four, like you can, you have so many options and that's really cool.
[00:33:43] Lauren Burke:
Yeah, data art is so interesting to me. I've seen a fair amount on Etsy. And like two of the examples kind of align with what you've mentioned. I've seen the ones where it's basically a poster of a movie, but it's the color, the main color, from every scene aligned into basically a giant poster. And then similarly for songs, it's basically the, I think it might be the pitch of the song over time.
And it's, it's just such an interesting way to present it. And it's so different from, I feel like a lot of what we think is Tableau data visualization where you're really focused on the interactivity. But if it's something you're printing out, you have to make it speak for itself without someone being able to interact from it and learn from it that way.
[00:34:22] Naomi Smulders:
Yeah. And it's also about it really goes back to the story you want to tell because at one point it's entertaining, but on the other hand, it's also like the kind of visualizations that if you're from far away, you already get what the main message is, but the closer you get, the more you see and the more intricate it gets. And it makes you question like, "What am I really seeing in this?"
Or like, "Okay, what does this mean?" And it makes you draw your own conclusions, but also really guides you to that conclusion. I think this is also an amazing skill. And I'm still trying to master it a bit better.
[00:34:57] Lauren Burke:
Yeah, I completely agree with that. So the final question I like to ask everyone, is there a resource that has helped you in your career that you think might help others who are listening?
[00:35:06] Naomi Smulders:
Yeah. The resource that I think mostly helped me is the community. Because I try to build a community anywhere. Like I said before, like DSconnect, I just started it because I didn't wanna lose these people. I wanted to keep 'em close so I could keep learning from them.
But I'm doing the same right now because I'm a data scientist in the experimentation field and I don't know a lot of people here. There are a lot of analytics people, but the data science part is still quite small. So I actually told some of my colleagues, who told some of their people in a conference and then I got into contact with this one guy who was also kind of like in the same thing that I am doing. But in a different company.
And then like, we had lunch and then he knew of another guy. So now we have this club of like four or five people and we just meet up every two months and talk about kind of like the struggles that we're facing at work, the problems that we're trying to solve and all those kind of things. And it's nice.
And I also had like one girl who joined because I just asked her on LinkedIn if she wanted to join sometime, cause she was near. And it can sound a bit scary to kind of like just approach someone out of the blue and be like, "Hey, do you wanna nerd out with me?"
But actually a lot of people are really open to it and like if they join once and then they figure out it's not for them, then it's also fine. But to kind of like have these people that are passionate about the same thing and try to unite them. That's my one resource because I can always call them whenever I'm in trouble, be like, "Hi, help."
[00:36:37] Lauren Burke:
That's awesome. I feel like most people want that community and want that community of people who are as excited about the things that they're excited about as they are. And for some, like the area you're in, it's fairly new. If that doesn't exist, someone has to start that. I think it's amazing that you've taken the initiative to know that you want that community of people who are as passionate about these things as you are, and then working to not only create it, but find the people that are going to help you build and expand that community over time.
[00:37:06] Naomi Smulders:
Yeah, it's, I don't know, it doesn't feel like work, you know? It's just, it's nice to surround yourself with people that have the same ideals. It's how you pick your friends most of the time as well. And it can be scary at first to approach new people about it, but I think people are open very quickly if they can talk about their passions or their awesome jobs.
[00:37:29] Lauren Burke:
That is so true. I completely agree with that. So how can our listeners keep up with you?
[00:37:34] Naomi Smulders:
Yeah, I don't have anything soon in the pipeline, but you can always just reach out on LinkedIn. And if you have any questions, LinkedIn is probably the easiest place to go. And I will try to answer them or put in a quick call to get there. And if I have any talks or if I'm ever on conferences again, I'll also post it on my LinkedIn and I'll let you know.
Yeah, and the blogs and stuff that I write are also coming on there.
[00:37:56] Lauren Burke:
Awesome. Well, thank you so much for joining us today, Naomi. It was so interesting to learn about process mining and about your background, the path you've taken, and some of the skills you've brought with you along the way.
[00:38:07] Naomi Smulders:
Yeah. Thank you very much. I really enjoyed being here.
An Analytics Community. Featuring Women. For Everyone.