- Data management
- Data Analytics
- Internet of things
- Artificial Intelligence
- Software Factory
- Customer Experience
- Data-driven ESG
- Case Studies
In the beginning, there was a data strategy, followed by migration to the cloud, the basics of data analytics for vectors, and corporate reporting, before the completion of the data lake, and everything is moving towards data democratization. "All the data models we had on-prem are now in the cloud, so we can scale and do everything exactly onclick. And because we have a data lake of one truth, we can now step towards data democratization and tell people: all the data is here, and you can click what you need," says Václav Dorazil, Head of Data at Eurowag, in the Adastra podcast.
- How to define a data strategy properly?
- How to successfully migrate from on-premise to the cloud?
- How do the cloud and data lake help companies?
- What are the benefits of data democratization?
Listen to the podcast (CZ)
Read the podcast as an interview
Ivana Karhanová: Define a data strategy, partner with senior business people across the organization, use data as a basis for decision making, migrate from on-prem to cloud, create a data lake of one truth for all analytics, and create a unified data layer. This is the mission of Václav Dorazil, Head of Data at Eurowag, a company that offers solutions to democratize the business of commercial road transport. Hi, Václav.
Václav Dorazil: Hi, Ivo.
Ivana Karhanová: You’ve been at Eurowag for a year. What if what I just listed is successfully behind you? And what do you still have to finish?
Václav Dorazil: I am perpetually dissatisfied, so nothing is ever behind me, but I think we have managed to lay a solid foundation. We have a data strategy, which means we know what we’re doing and the vision for what the data should bring to the company.
We know how we want to work and what we want to offer the company, and we have a prototype MBP, and we are finalizing the data lake in production. We’ve laid the groundwork for data analytics for the vectors and the foundations of enterprise reporting.
So I think we’re on a good path, we have a lot of challenges ahead of us, but most importantly, we’ve laid a solid foundation to grow, be dynamic, and be a solid partner for the business next year.
Ivana Karhanová: You have migrated from on-prem to the cloud. You defined your data strategy at the beginning. Eurowag didn’t have one before?
Václav Dorazil: We had a lot of ideas and visions, but usually, it goes to the data team, and you can tell if the company is serious about data. The moment they set up a data team and say: Here’s the data team, and they’re going to take care of it now.
It’s important to understand that the data team always has a lot of stakeholders and customers in the company. And everybody expects something a little bit different from the data team. So, for example, senior management wants some KPIs and reports, and Legal and Compliance want us to do something with the data to follow some data.
Data and engineering teams want us to help them with data quality, to work with everything somehow, and everybody wants a little different.
Ivana Karhanová: What does IT want? You left that out.
Václav Dorazil: IT wants a lot of help to do things right, to do them on sensible technologies, to have support in scalability, and to do things sensibly, not to be one step behind the business, but to have time to be ready for what the business wants.
There are a lot of general talks in IT about data quality, so real production teams just want to help with data quality. What is that? From such an abstract term to something tangible, how do we measure it, how do we help with it, and how do we like to improve it. I think that’s a daily challenge for engineering teams.
Ivana Karhanová: You said we defined a data statement initially. Why did you do that in the first place? What was the initial impetus when you came in and said: Do we need this?
Václav Dorazil: I guess transparency. And just because there are a lot of stakeholders and a lot of expectations, so the data statement is nothing complicated. But, still, it’s about listening to all the parties, what they want, what they’re struggling with, what’s the problem they have with data and what they’re solving, and then offering it transparently to everybody and saying: Look, we’ve listened to what you all expect us to do and here’s some A4 of what we should do then because sometimes it can be contradictory.
Somebody wants something quickly, but maybe the compliance leader wants, hey, here are the exact requirements. We must comply with some GDPR data and treat it specially. So the problem statement tells us that this is expected of us. So we can prioritize it in some way and tell the company: We can’t solve everything, but next year we’re going to focus on points 1, 2, 3, 4.
Ivana Karhanová: How long did it take you to put this together?
Václav Dorazil: One quarter, I would say. Usually, it takes a quarter. It’s a lot of talking, listening, and that’s also what accompanies the data. The data is not just technology. It’s not some zeroes, ones somewhere in databases. Still, it’s always putting together the business, understanding what’s expected, and then you need to somehow translate that into something we should be doing. So one quarter, I think it’s going to take.
Ivana Karhanová: Eurowag, like many other companies, suffered because data was locked up in data rooms. How did the owners of the data silos react when you told them that now the data was going to be handled differently and that suddenly their data would be available for others to see and work with?
Václav Dorazil: Well, that was part of the problem statement because one of the big priorities or the big problems that bubbled up for us was that the data was in silos. The impact then on the business is simple.
When somebody wants some insight per customer or maybe a vehicle, which is an important entity for us, to combine data from different systems as the business grows and scales, it’s just not easy, and it takes time to get to that. So this was an important part of the problem statement, and then individual teams, engineering teams, or product teams responded very well to that.
Why? They’re also swamped with operational stuff, and somebody’s always asking them to do something, do some one-off data dumps, and combine it through Excel. And the fact that we tell them: This is our vision, we’re going to have a centralized data warehouse, where we’re going to connect all the data, and that’s where it’s going to allow us to respond very quickly to the demands of the business.
So I think that’s what will convince those teams to work with us and say: Oh, we’ll then be relieved in the day-to-day operations, and we won’t have to deal with so many ad hoc requests that we have now.
Ivana Karhanová: Did it take you a lot of work to convince them? Because we often hear from customers all sorts of reasons why it’s not going to be easy and why it’s not going to be easy to convince so-and-so to release the data.
Václav Dorazil: That’s a good question. I try to be realistic and pragmatic. And pointing people exactly to these real issues or to solve real problems. Those teams suddenly see that we’ve built two, three, or four reports.
When there are, for example, close financial issues, we’re able to go in and help them very easily look at where what data is, how it’s flowing and what the meaning of it is, what the status of, for example, a transaction is, whether it’s already booked.
That’s a standard problem, what’s posted, what’s pending, what’s in what status. Having it all in one place allows us to respond very quickly. And that quick resolution of those issues ultimately convinces those people far better than some directive command from management that we just have to put it in one place.
Ivana Karhanová: So you decided to go to the cloud. There was a time when the cloud was presented as a cheaper solution, which is not exactly the case. What was the real reason for you to go to the cloud, and what was the cost associated with that?
Václav Dorazil: Good question. I think the cloud is unavoidable for companies of a certain size and above. From my perspective, the business is evolving, and I’ve already mentioned some booked transactions. The standard first partner of data teams is usually finance or finance teams that are working, doing some closings and reports, and that’s how it starts in most companies.
Ivana Karhanová: Logically, shouldn’t it be a business that asks where are our reserves, what’s working, and what’s not working?
Václav Dorazil: That will come mostly in the second stages. Those companies in the beginning – and I think Eurowag is very similar – are going in that direction. Eurowag has come to that point recently, but even the basic or financial requirements, the basic corporate reporting, stop scaling from a certain size. Eurowag is typically a company with a big acquisition appetite.
That means that even those basic reporting requirements require data consolidation across different entities, whether in Portugal, Spain, Italy, or Poland. And while that’s possible in that on-prem architecture, it’s harder and takes time.
Ivana Karhanová: I have to send the data somewhere, somebody has to put it in, somebody has to collate it so that it’s the same as the others, and then it just starts working?
Václav Dorazil: The right perspective, which kicks it up a lot in the end, is always when the company starts thinking more business-wise, not just financially and controlling-wise. Eurowag, after all, is moving there, too, from some of the early days. From commodity business, we are now looking much more at how to digitize and set up some platforms, load matching, aggregate truck information from some thematic data, and help them and optimize routes.
And that’s just very, very hard to do in the on-prem. That’s huge data you just need to have somewhere and have it accessible onclick. And that’s what the cloud will allow.
Ivana Karhanová: So, in the case of Eurowag, it was not a question of finances, whether it was cheaper to stay on-prem or go to the cloud, but a question of further development?
Václav Dorazil: Further development and business development. Data and technology, in my opinion, should not invent what they want to do themselves because there are always so many possibilities, and we can spend ten years developing anything, but what we are promising from the data lake and the cloud is to solve the first problem or one of the pressing problems.
That is to be able to plug and play really, very quickly. There will be new data sources, new acquisitions, and new systems. And what we’re building simultaneously as the data varnish will help us respond to that very quickly.
And then the ability to – to put it simply – chop data models like Bata drills, that’s important for the business because we’re getting into a situation with the cloud and with the data varnish where we can be much more agile. So we don’t have to wait half a year and explain to the business, “But we now have to do and get the data from A to B and then combine it and then play with it and clean it up somehow.
Ivana Karhanová: And then they’ll be obsolete.
Václav Dorazil: Exactly. It’s just too late, and we’re trying to think long-term and be one step forward, not one step back.
Ivana Karhanová: And you also told me before the shoot that now you can finally do data science, whereas before, you were doing data engineering. If you were to explain this to the top management of the biggest companies, what would be the difference for you?
Václav Dorazil: For me, it’s simple. First, we must learn to walk, and then we can run. I understand that everybody wants AI, machine learning, and more. It’s nice, but you must have some basic data background and honor the culture.
There are generally three basic roles and responsibilities in data: the first is the data analyst, the second is the data engineer, and the third is the data scientist. As I said at the beginning, we’ve laid some fundamentals; we have the data in one pile. We can combine it, and now only then can we let the data scientists come in and say: Hey, we can play with this, we can make predictions here, we can do data model here, we can help you here, but in the reality of everyday life.
Suppose we had hired a team of five data scientists a year ago, for example. In that case, I think in one or two months, they’re all gone because they’d have to do data engineering, they’d have to migrate data from A to B, they’d have to worry about some quality and they couldn’t do data science.
Ivana Karhanová: You’re trying to build a data lake for Single Point of Truth. That means that everybody should see the same data and be able to draw at least similar conclusions if possible. How does that work in practice now?
Václav Dorazil: It works. It’s a lot about interpretation because the Single Point of Truth is very important. From my personal experience, I was taught mainly by Kiwis. I’m a big believer in data modeling. Data modeling, data model, and Single Points of Truth are where all data people find that data.
There are a lot of talks now about data democratization, about enabling all people in the company to access data. That’s a cool concept, and we’re also moving toward that. But we need to work hard on it first, sit down with everyone, clarify what a customer means, what an active customer means, and what it means when a customer churns. Those are simple definitions.
And then, we need to transform those definitions into a unified data model that gives access to all the people in the company and avoid everyone having 100 or 200 Excel sheets combining different data that flows in different ways, and no one can manage.
That way, we have one source of truth, and now we are able and willing to step towards data democratization and say to people: you can find all the data here. You don’t have to be super techy. We’ll let you click on what you need. And it’s so well done and well described that you can do it yourself, and you won’t need a data team.
Ivana Karhanová: How long did you transition to the cloud take?
Václav Dorazil: About a year. I think a year, I think we started flirting with it at the beginning of this year. Now we’re nearing the end and finishing production, so I would say a year.
Ivana Karhanová: Moving to the cloud is lengthy, full of stumbling blocks, and often expensive for many large companies. What is your recipe for a successful migration?
Václav Dorazil: Do it with a vision of what it should look like in the future. For us, it’s important not only to look at the cloud but also at the overall data architecture and the ingestion layer.
It’s always important to maintain some continuity. So we’re going to copy that solution, even the one we have now, into the cloud because nobody can afford to suddenly come in and say: This and this will stop working.
Ivana Karhanová: So you did a lift and shift of the architecture. As it was, you just took it and threw it on the cloud.
Václav Dorazil: I wouldn’t say the whole architecture. I would say the targeting of all the data models that we had in the on-prem, we moved to the cloud. This gave us a little bit of performance relief. And now we tell the company that we have everything in the cloud, we can scale, and we can do everything onclick accurately.
But now it’s important to take that and review what’s there then. Companies should always look at that, if only for the possibility of rebooting and discarding some of the technology debt that inevitably always accumulates. So next year, we already know this. We will be doing a review and simplification of what’s there.
Ivana Karhanová: Now you can easily find that a lot of the things there are broken or expensive from a cloud perspective. And maybe you’re migrating some data three or four times back and forth on top of that.
Václav Dorazil: Exactly. I believe that data teams should be data-driven and always to all data models. We should know how much ETL, how long it’s been running, and how much it costs the company.
Then when someone comes and asks how much corporate reporting costs or how much some vehicle or customer analytics cost, that data team should be able to tell you: It’s this part of our data paint, data warehouse, or dedicated SQL pool, this is how long it’s been running, this is what it’s costing us.
We want to turn it on, turn it off, and give that room for discussion based on the data and the data inputs. At the same time, with that reorganization or restructuring, it’s always possible to throw away some part that we’ve just historically developed and simplify it.
Ivana Karhanová: You now know how much is costing you which data and how much is costing you which report?
Václav Dorazil: Historically no, newly I hope so. I don’t know if we will be able to go up to the level of reports, but we will be able to go up to the level of some areas or series of reports.
Ivana Karhanová: In your words, Eurowag has also moved forward technologically by about fifteen years in the one year that it has migrated to the cloud. What’s next? Lift and shift you have done. You are reviewing processes. Are you also reviewing KPIs?
Václav Dorazil: We are definitely reviewing KPIs. It’s important to have a top to bottom approach to KPIs and all reporting. It means that some metrics are wanted by senior management. These are the main indicators of the company, and what is the net revenue? Then there is a second level, and let’s say, the business unit.
We have energy and tolls and other things that also have their metrics, and companies need to know how the business unit metrics fit into the enterprise-level metrics. Then, finally, the last basic level is product metrics. Every company has some product teams, some service, self-care, customer service, or some business. And that team is working on some products, and they need to know how their products fit into the business unity and enterprise metrics.
So I think that’s an important task, just the interconnectedness of those metrics so that everybody knows how they fit and contribute to the greater whole. So that’s what we’re going to see, and I hope those will be interesting things for the business next year.
Ivana Karhanová: Many companies are looking for a Chief Data Officer. How do you think you know a good CDO?
Václav Dorazil: How do you know? Simply by knowing what they are talking about.
Ivana Karhanová: Many people can speak well, and you can’t tell at first glance that they are slightly detached from the practice.
Václav Dorazil: That’s right, and I think that any good manager should always listen for a while and then come up with a plan that he or she can implement, not making promises, not hanging on the nose, not promising big visions. It’s good to have those visions and to direct them. We have them in Eurowag, and I think we have them well described.
We have data visualization in the data strategy, but now we have to do the hard work. We have migrated to the cloud in a year, which is a super achievement, and I believe we are a good partner for the business. That’s for others to judge, and ultimately it’s good to ask the people, the individuals working with the data, if where we’re going is good and if we’ve done something after this year that’s helped them.
Ivana Karhanová: I was asking you what was the reason why you went to the cloud as Eurowag. You gave me a very simple answer: Because I decided to. And for me, that’s a super simple explanation. But it’s so simple even in practice that the Chief Data Officer will say: Are we going to the cloud?
Václav Dorazil: It is. It’s always the company culture that is important. But again, we have a data team, and they have some responsibility, but they should also have some authority and say what they are doing and why they are doing it.
That means we’ve listened, and we know what the company expects from us, but we’re the ones who have the authority to say the cloud is the solution to this problem. So we know why we’re going into the cloud and what we expect from it. We know it’s not free, but at the same time, we also know what it will bring us.
Ivana Karhanová: Do you think there are enough people like you in the market who will just come and say: Guys and gals, the cloud is the solution, so let’s go for it?
Václav Dorazil: I don’t know. Let others judge.
Ivana Karhanová: But do you know what the market looks like?
Václav Dorazil: Yes, I know what the market looks like. I think we are really in a phase where we talk and write home a little, as my grandmother used to say. So I like to write home, and I like to be realistic. And I’ll leave it up to other people and companies to approach it the way they want to.
But I think it’s good that there are Chief Data Officer positions that companies are considering along those lines. The closer that position is to the executive and the board, the better it is for that company. And it’s always a sign that the company is serious about data.
Ivana Karhanová: Data has now become an asset for Eurowag. What does that mean? When we translate that into practice and fulfilling the vision that Eurowag has?
Václav Dorazil: We look at data in a truly product-oriented way. Well, I’m a person who usually tries to look at it as a product. The data team is making a product, it’s a data model, and it’s something that has some customers, and at the same time, it costs us something.
And that’s the same way we’re trying to look at it now. If we have some business domains and we’re going down the path of maybe DFF or load matching. That’s how Eurowag is trying to help its customers, its partners, and at the same time, the environment.
Ivana Karhanová: You have mentioned DFF several times. Let’s explain this term.
Václav Dorazil: It’s Digital Freight Forwarding, basically the ultimate goal. Simply put, it’s a combination of some fleet management systems: truck management. DFF, as Digital Freight Forwarding, means the forwarding of goods. And the ideal golden grail is to help partners optimize truck routes so that they are as close as possible to other goods and transportation.
Ivana Karhanová: It reminds me a little bit – and correct me if it doesn’t – of the platform that Liftago recently launched, which wants to implement The Same Day Delivery by offering a marketplace with spot shipping to contracted e-tailers and shippers. Something similar should be created on your side for trucks.
Václav Dorazil: Simplistically, of course, that’s true, but the important thing is not only the benefit for the shippers per se, but now the topic of emissions and other things is very hot. According to various studies, up to 20% of trucks run empty.
Maybe every percentage point that we help to optimize the journey so that the truck does not run empty but actually carries some goods from A to B and then finds another one from C to D as quickly as possible not only has an impact on the hauler, who will pay less for petrol. So it will be much faster and more efficient, but we will of course also help the environment.
Ivana Karhanová: Does that mean those who order transport should know that they now have the option of adding their goods to a truck that can do so and is going in the same direction?
Václav Dorazil: Exactly.
Ivana Karhanová: How long will it take you?
Václav Dorazil: I honestly don’t know. But we will work agilely, and I believe it will be one of our priorities next year. So we’ll see where we get to.
Ivana Karhanová: Without the cloud, you couldn’t implement this?
Václav Dorazil: No.
Ivana Karhanová: Says Václav Dorazil, Head of Data at Eurowag. Thank you for coming into the studio and sometimes on another topic – and I think it comes down to the platform – to be heard.
Václav Dorazil: Thank you, bye.