The swiss knife approach to data team at EQOM Group

Tarush Aggarwal (00:00):
Hey, everyone. Welcome to another episode of People of Data, where we get to highlight the wonderful people of data. These are our data leaders from various industries, geographies, and disciplines. Discover how companies are using data, their achievements, and the challenges they face. Today on the show, we have Irina Ioana Brudaru. Irina is a modern Swiss knife with more than 15 years of experience in computer engineering with a research background. Irina spent more than six years at Google. She's now leading and working with different companies in the EQOM Group. She's also passionate about mentorship and leadership with a strong focus on supporting women in tech. Welcome to the show, Irina.

‍

Irina Ioana Brudaru (00:43):
Thank you for having me, Tarush. It's a pleasure. Yeah, where shall we start?

‍

Tarush Aggarwal (00:47):
Let's get right into it. We'd love to find out what does the EQOM Group do? What do you guys do? How do you make money? And how does data fit into all of this?

‍

Irina Ioana Brudaru (00:55):
It's... yeah! I have to say that most people say, "What is this EQOM Group?" I've never heard of a big company like this in Europe. And then they look, and they're like, that makes sense. So EQOM Group is a sexual wellness group of companies with, let's say, daughter companies in Germany, France, the Benelux area, the Nordics as well, the south of Europe. So we have something a little bit everywhere, but our strongest presence is in the, let's say, from France upwards geographically. Sexual wellness means a lot of things—means toys, means education, means classes, understanding, and being. I would say working in this field makes the people very open-minded because there's, you know, just imagine that if you don't have a sense of humor and you're, you know, maybe a girl who can't do this. No such thing exists here.

‍

Tarush Aggarwal (02:08):
Mm-hmm.

‍

Irina Ioana Brudaru (02:16):
It's very nice. Our team, so to give a bit of context, in my previous roles I was a Head of Data, but I decided to take a little bit of a step back, and my current boss, Tom, he's the Head of Data. Though I would say at the group level, it's just a completely different beast to be a Head of Data for a whole group.

‍

Tarush Aggarwal (02:41):
Awesome.

‍

Irina Ioana Brudaru (03:05):
He snapped me off the market when he saw my profile and said, "There are so many cool things to work on at EQOM. Please join." And so I did. It's really fun. So we are a team. It's also, if I would say, this is a very unique way of working in our data team that I have never seen in any other data team before. We're all almost working like independent consultants. So our boss doesn't tell us what to do. Somebody needs an analysis, go and do it. It's more like we are all quite senior in our roles and have a little bit of multiple areas of expertise. So we're all a little bit of a Swiss army knife with different tools inside it, and we can cover a variety of data areas between each other.

And this is also why our boss does not give us work. We know we go to all the meetings that are important and we know what the gaps are and what needs to be done. So it's such a high level of accountability, trust, and responsibility. It's really, really nice. The projects are more like consultancy projects because you go in, you do something, and then you go out and do something for someone else. Now, because this is a group that has in recent years purchased new companies, ingesting the data and leveling the data maturity at the group level is not something that is easy. It is reflected in our current ERP and database projects and building the data. I personally work on really nice ML and data science problems. So in this role, I'm doing something I haven't done before, which is media mix modeling for understanding the impact of marketing spend. It's an art in its own right. It's a very big beast, but we're using a tool that makes iterations easier and the results faster to use. I'm also using causal inference for measurement of all different things. You have different ways of measuring things and A/B testing is the most classic—this is for when you do not have as many signals. How a poster in a city or number of posters have impacted sales or how spreading flyers in a certain neighborhood impacted sales. So you can build test and control from these things.

Yeah, I like to work in this little sandwich between data and business. I'm motivated by making a financial impact, to be honest. And, of course, user experience and whatnot, but financial health for a company comes also from user experience and health on the website.

‍

Tarush Aggarwal (05:46):
Yeah.

‍

Irina Ioana Brudaru (05:46):
Yeah. Yeah.

‍

Tarush Aggarwal (05:46):
That makes a ton of sense. So kind of at a macro level, you have multiple different businesses. As you mentioned, you have trainings, you have courses, you have an e-commerce store, you have all of these different business lines. And that's essentially how the EQOM Group operates. How does data play a role in helping the e-commerce group fundamentally work across these multiple businesses?

‍

Irina Ioana Brudaru (06:19):
There are areas of the business that are very similar in the way that the reporting is standardized. Maybe some of the companies are a little bit higher in data maturity, and they will say, "Feel free to try something new on us," and if it makes money, then we'll do it. And then they become the inspiration for the rest of the group if they are leading specific new topics or new ideas.

There is always the red line of EBITDA. I'm a huge fan of EBITDA reporting and of unit economics. And all in all, I mean, all companies are separated not only geographically but to a degree also on culture and communication. So yes, these also are aligned through the data and through the story that the data tells.

Currently, for example, we're looking into CRM, and there are a couple of companies that have worked with CRM tools for much longer than others, which are newer. So instead of sitting in your own little bubble and saying, "I only know the German market, and that's the only thing I can do," no, go and ask your friends at another company. We're not competitors; we are working together. So teach us for free how you use this tool. Why should I pay a...

‍

Tarush Aggarwal (07:08):
Yeah.

‍

Irina Ioana Brudaru (07:35):
A provider teaches us all the tricks of breaking these cultural barriers with data. I feel like it's part of the job as well.

‍

Tarush Aggarwal (07:44):
Yeah.

‍

Tarush Aggarwal (07:44):
I love it. It seems like the culture is very open and sort of very collaborative across these different entities which you have. If you just zoom into the footprint of the company, how many employees do you have across all of the different groups? How many employees are in the data space?

‍

Irina Ioana Brudaru (08:01):
I am.

So the companies, all employees summed up, I think, are about 500 people overall. I do not know everyone. I work in our team; we work remotely. I'm in Germany, but the mothership is in the Netherlands in Amsterdam, where also the storage is. Yeah, that's a lot of people in a lot of countries. So we also have offices in France. We also have offices in Germany, Netherlands, and think Belgium is up for discussion and a couple of others. So yeah, you can, for example, work for the Nordics but be in Spain. It's okay if your boss agrees with your performance and the setup is great.

‍

Tarush Aggarwal (08:37):
Beautiful.

‍

Irina Ioana Brudaru (08:47):
A couple of years ago, data was considered a shared service. So everything was brought to the mothership. Our boss belongs to the mothership and yeah, he's leading everything for all of the companies. Our data team has five or six people at the moment, which is not a lot at all. But if you remember that I told you we work as very senior consultants, that's also why you understand why we're so impactful. That doesn't mean that it's enough people, but we know how to work smart, prioritize. And yeah, our team is located remote as well. This is a perk of our company. And actually for most people, it's a huge perk to be able to work remotely from anywhere in Europe.

‍

Tarush Aggarwal (09:21):
It never is.

‍

Irina Ioana Brudaru (09:27):
Yeah.

‍

Tarush Aggarwal (09:27):
Yeah.

‍

Irina Ioana Brudaru (09:38):
We also have Databricks, and I love Databricks so much. It's the first time I got to try it. It's so natural to find what you need and to build things. I also really like that they have the SQL notebooks similar to Python notebooks. So you can run, test, and connect, and so on. It's hugely fun and fast.

‍

Tarush Aggarwal (10:08):
Yep.

‍

Irina Ioana Brudaru (10:05):
Yes. So would say given the different levels of digital maturity and data literacy in the company, you need these tools because finding somebody to have all of these skills in one is harder. Having such tools allows you to iterate faster and move forward faster.

‍

Tarush Aggarwal (10:14):
Yeah.

‍

Irina Ioana Brudaru (10:33):
And you wanted to ask what the stack is. We have Databricks, and I love Databricks so much. It's the first time I got to try it. It's so natural to find what you need and to build things. I also really like that they have the SQL notebooks similar to Python notebooks. So you can run, test, and connect, and so on. It's hugely fun and fast.

‍

Tarush Aggarwal (10:48):
Yep. Yeah. Yeah.

How do you get data into Databricks?

‍

Irina Ioana Brudaru (11:04):
Through very many ways, through APIs, through tools, through import from some legacy system, expensive tools, but there's also a feature in Databricks coming up that would replace ingestion tools.

‍

Tarush Aggarwal (11:13):
Yeah.

‍

Irina Ioana Brudaru (11:20):
For dashboards, we're just using a classic tool. It's not like there are very many tools. We're not there with the semantic layer and Looker or anything else. It's just early. Of course, the classic Google marketing stack, retargeting stack, and so on.

‍

Tarush Aggarwal (11:34):
Yeah, so when you mentioned BI, what's the current BI tool across the company?

‍

Irina Ioana Brudaru (11:39):
Power BI, of course. Yeah, I know. You can even have things like little Python plugins or R plugins and bring that fancy stuff into it. You can do anomaly detection directly in Power BI on the charts. It's very nice.

‍

Tarush Aggarwal (11:41):
Power BI. Yeah, I think Power BI is a fantastic tool. Makes sense.

‍

Irina Ioana Brudaru (11:47):
Yes. Exactly. So we're using Power BI as our main BI tool. And of course, we use DBT because otherwise, we mentioned orchestration. Of course, we have to. Otherwise, orchestration from all of the...

‍

Tarush Aggarwal (12:25):
Yeah, and are you using DBT Cloud or are you using DBT Core?

‍

Irina Ioana Brudaru (12:28):
I don't want to say something stupid, but I think it's Cloud. I'm not sure because I'm not touching the other projects.

‍

Tarush Aggarwal (12:32):
Yeah, got it. No, it makes perfect sense.

‍

Tarush Aggarwal (12:32):
That's awesome. And a stack like that really allows you to do a lot with a smaller team because you're not managing and maintaining 10 different tools to go do it. You're already leveraging the power of Databricks. So that makes a ton of sense.

‍

Irina Ioana Brudaru (12:46):
Yes. So I would say, given the different levels of digital maturity and data literacy in the company, you need these tools because finding somebody to have all of these skills in one is harder. Having such tools allows you to iterate faster and move forward faster.

‍

Tarush Aggarwal (13:05):
Yeah.

‍

Irina Ioana Brudaru (13:25):
I would say data-driven organization and making data an integral part of every step, giving priority to personalization is what I'm...

‍

Tarush Aggarwal (13:53):
What's one example, like in the, like what's one real-life example where, you know, what the world looked like before this and like post this project, like what did that look like?

‍

Irina Ioana Brudaru (14:05):
Okay, if there is a person in the till who wanted to have the data, understand like person X comes in and says, "I have been using Virgin Media for the last 10 years, and I want this." And if they press in five minutes, they get the dashboard of what all the products they have used. And AI generating a recommendation as well, telling like, "Hello, Mr. X, you have been with us for these many years, and this is what your usage is. Based on your usage, this is our recommendation."

‍

Tarush Aggarwal (14:42):
Got it. So a recommendation engine is automatically fed in with all of the customer's data, like previous bills and how long they've been a customer. And how are you sort of running this sort of decision engine? What's the compute behind it?

‍

Irina Ioana Brudaru (14:59):
It is the data intelligence, right? Like how you are having the data and what all the information available plus adding the data science models to do the predictions. Yeah.

‍

Tarush Aggarwal (15:08):
Sure. Is this based on generative AI, or is this more like traditional data science algorithms?

‍

Irina Ioana Brudaru (15:17):
I would say it's a combination. It is not like one or the other. It depends on which data product you are referring to as...

‍

Tarush Aggarwal (15:28):
And are you using any of the usual suspects when it comes to Gen AI, like OpenAI, Anthropic, or is this more an in-house deployment?

‍

Irina Ioana Brudaru (15:40):
It's again a combination, Tarush. The scale of the organization and the data we have, especially the network data, is in trillions and millions of data points. So it's not like a straightforward target, like you have 100,000s of records, let's analyze it and build a model. It's about what kind of category of data it is and using OpenAI and Gen AI, and also building custom tools to get the result we want based on the business needs.

‍

Tarush Aggarwal (16:16):
Got it. What is one of the challenges that you face today which you're actively working on solving?

‍

Irina Ioana Brudaru (16:24):
Skills gap.

‍

Tarush Aggarwal (16:26):
I would love...

‍

Irina Ioana Brudaru (16:28):
For the university. You need to understand what skills are required for the persons to understand the data.

‍

Tarush Aggarwal (16:35):
Got it. And do you see this across the entire company? Are there particular departments which you want to empower before others? How do you basically roll out a university across 70,000 full-time employees?

‍

Irina Ioana Brudaru (16:50):
I would say first the data people have to be ready before the, I mean, the 500 people have to be skilled. But again, there is no first-come-first-basis as such. We would like to have a parallel plan as well, running some business people, also empowering them to understand the data while they are having a vote for doing this. And the data people also getting enhanced with their skills and all that would help both happen at the same time.

‍

Tarush Aggarwal (17:24):
Got it. And very practically speaking, how would you go about solving the skills gap problem? Make sense at a strategic level that you have Data University, and this is an area which you're thinking about on a practical basis. What do you measure in terms of what's missing and how do you basically go impact that? What impact is this potentially going to...

‍

Irina Ioana Brudaru (17:45):
You're asking me to share my strategy plans to you now. Anyways, the way we do it is technical assessments will be done based on how many people are actually needed and what their skill sets are. And also measure the skills, how much they have and what they're missing at the moment. And create a learning path for them to actually build their career and build their data skills. And give them the badges, whatever is required from the university side. So depending on the number of badges they earn, they will reach the levels. And that will go on based on their performance and based on their badges that will combine together and give the plan.

‍

Tarush Aggarwal (18:27):
Got it. And how do you see this impact the organization in the long run? Like obviously, data literacy makes sense on paper, but you know, what do you expect to be the practical implications of sort of self-service and data literacy?

‍

Irina Ioana Brudaru (18:42):
I would say everyone will become independent, empowered, and often they even realize how much data could do. The power of the data will be more evident, and it will become a more data-driven organization. That means we could do more within less time. That means we get more customer satisfaction. Customers will be happy, and we are happy.

‍

Tarush Aggarwal (19:00):
Amazing.

‍

Tarush Aggarwal (19:00):
Bhagya, thank you so much for being on the show and sharing your wisdom with us.

‍

Irina Ioana Brudaru (19:09):
Thank you, Tarush. Thanks a lot.