Data science in Business
This week, we are overjoyed to present an interview with one of Aegon’s Data Scientists, Rick van Dael. Our members of SECTOR had the opportunity to hold a lengthy conversation with him about topics from his study in Maastricht, to the energizing and young culture at Aegon, to Aegon’s future in respect to their goals and ambitions, to how one becomes a Data Scientist at Aegon, and much more. This article aims to give a good picture of how it is to work at Aegon, and what opportunities the insurance giant represents for newly minted econometricians, we hope you enjoy it!
What is your background?
I studied Econometrics and Operations Research in Maastricht, where I completed both my bachelor and master. I have been a working student at Willis Tower Watson in Eindhoven, I did an internship at Accenture, and finally I ended up as a Data Science trainee at Aegon.
How did you come to the traineeship at Aegon?
When I was a working student at Willis Tower Watson, my work was very actuarially focused, which I enjoyed, but I was uncertain if I wanted to become an actuariat. It was at the same time that Data Science started to take off, and it seemed like an interesting sector to be involved in, but that was not possible at Willis Tower Watson. It was possible at Accenture, but that was in Heerlen, and I did really want to move to the Randstad. In addition, I was unsure if I wanted to be a consultant. Thus, I started searching for companies where I could work as a Data Scientist, and I eventually found Aegon.
What made you choose Aegon?
The international character of the company. To give some context, the traineeship consists of modules facilitated by MICompany, a data science consultancy that also offers courses in data science, where employees of KPN, Aegon, Postcode loterij and a host of other companies can join. The programme therefore is relatively comparable to that of other companies, but because of Aegon’s international reach, you have a number of options to do things internationally. One such opportunity is to work abroad or work at one of Aegon’s daughter corporations. During my study, I went abroad to Madrid and Vancouver, and I enjoyed myself immensely, by which I wanted to seek out a company with an international character. I quickly found that at Aegon, where I worked for half a year in Brazil, at the office in Rio. Aegon is quite unique in this, as other companies with similar training programmes often either were entirely based in the Netherlands, or did not offer exchanges with their foreign subsidiaries.
What was the reason you decided to forgo an academic career to work in the business sector?
From my point of view, a PhD was a relatively long commitment, in particular because I wanted to do a PhD under a professor who I had a personal connection with. This was possible in Maastricht, but I had already spent 4 year studying there, and I would rather not spend another 4 years there, because, while it is a wonderful city, it is quite removed from the rest of the Netherlands. I did want to do a PhD in Amsterdam, but I would have to do a new master, and I did not know the professors, which seemed like a relatively large risk to me, because I would be tethered to a single person. In contrast, you always work in a larger team in a company, which does not have that same risk. I did not necessarily have academic aspirations, so a PhD would be more for my CV, which did not seem like the proper motivation to do a PhD to me.
A PhD usually gives rise to a more research/technical focused position, do you think that the Data Science development programme at Aegon also allows for this?
Certainly, that is possible. However, I personally discovered through the traineeship that I enjoyed the soft side of business, which Aegon also encourages and helps you develop. The Data Science department are internal consultants to Aegon, where we do not work for a single department, but we are a central part of the organisation. We have a Chief Data Officer (CDO) who we report to, and she in turn reports directly to the CEO. Essentially, we are another branch in the organisation chart, and through that we are deployed throughout the company. In essence, you have two sides to your work here: one the one hand you have the stakeholder and product management side, where you coordinate the project and present the results to your client, and you have the more technical side, involved in coding, building the model, the cleaning and processing of data, etc. If you consider those two sides, you can really specialise in being able to build a well working and incredibly complex model, but you can also focus a lot more on the relationship to the client, both of which are interesting. The soft side of the work can be particularly interesting, as it can create a nice break from all the technical stuff you have been going through for 4 years.
How was the switch from an academic to a business oriented training programme?
It was a very different experience. I originally doubted whether or not I wanted to do a PhD, which in the end I decided to forgo that opportunity. During your education as an econometrician, you have a large number of theoretical courses, but the real world is often not as easily and nicely modelled by the models you encounter in university.
What is the main programming language used at Aegon?
The main programming language is shifting from R to Python at the moment, and I know this, because I did everything in R during my traineeship, but the traineeships that came after did everything in Python. Besides that, you will always need SQL to extract the data you want so that you can use different models on it in Python or R. I do not know where the sudden shift came from, but I do think it is a pity, as I am a real fan of R, and now I have to learn quite a few things in Python. It is not necessarily a challenge to make the switch, but I am staying with R for now, mainly because I am better versed in it. I think the switch mainly comes from the fact that Python is more popular and more widely used, by which it would be a good idea to join this trend, because you can see it as a larger part of the future. However, there are pros and cons for every language, but there are more than enough people still programming in R at Aegon, and as long as I can work together with them I am more than fine.
What is the reaction from friends and family when you tell them about your work?
That is a good question. People who did not study econometrics often question what I actually do all day as a Data Scientist, where properly explaining it remains a challenge. I often hear from others that they think my employer is really cool, and that it is a fun job with a young and fun team, where lots of activities and drinks are organised, which makes it a lot of fun. I moved to Den Haag from Maastricht, without knowing a whole lot of people, so it was quite important to me to have a cohesive and close-knit team with whom I could also go out to the bars with. This makes for a rather special culture within Aegon, as people are connected through more than just their function. Besides that, my exchange to Rio is also seen as something incredibly cool, where friends who are doctors could not imagine that from their profession, which makes the international character of both Aegon and the job really stick out.
What do you think you prefer, building an extremely technical model or fitting a simpler one to the needs of the client?
I am a fan of both. For example, during my project in Rio I could basically go crazy with the models one could deploy, as we wanted to make a cross-sell model for insurance products, where we wanted to calculate the cross-sell propensity per customer. The goal was to gauge which customers would buy product B if they were already subscribed to product A, which would give us an ordering of which customers should be called first for a sales talk. It was an extremely interesting project, with a lot of freedom when it came to modelling this cross-sell behaviour. It was also quite fun to go through a number of papers to get caught up on what developments were made in the machine learning field, through which we knew how to best predict the cross-sell tendency of customers.
Do you often deal with large and complex data sets such as cross-sell data?
Indeed, especially in the Netherlands, with the regulation surrounding privacy and the ever watchful eyes of financial regulators such as the DNB, AFM, and others, which can be quite a challenge. It makes working with the data exciting in itself, but we do have to deal with these regulations, by which we are closely tethered to the privacy department, with whom we often check if certain methods and usage of data is allowed. For example, can we use the data if we hash it, make the names unrecognisable, or remove certain attributes from the data set, i.e. personal details which we are not allowed to use by regulation.
Do the models often produce surprising results?
Yes, that does happen. That is our function within the organisation, through which we are trying to make Aegon more fact-based. Aegon is a rather large company, where a lot of people have differing opinions, some of which are no more than a gut feeling or a refusal to part with the status quo. Thus, as a data scientist, you can really use data to provide insights on the real state of the business, by which you can create data driven insights that go against the grain. For example, there was the perception that our elderly customers were not very digitally savvy, with the prevailing idea being that they did not get to our website, did not have a MyAegon login, or did not primarily use their email address to contact us. However, this perception turned out to be quite false. This age group will be retiring relatively soon, and so they are very driven to manage their pension online, from the comfort of their couch. This group were shown to have a MyAegon account more often, as it was more relevant to them when trying to take care of their retirement, while something in their thirties would hardly spare a second thought on getting a MyAegon for their pension at that stage in their life. The results were quite surprising, but they provided valuable insights into our customer database, which makes subverting such a gut feeling and employing a facts-based approach all the more fun.
What are Aegon’s core tenets?
Aegon group, the central organisation overseeing everything in the world connected to Aegon, have recently released a new global company wide purpose. Just in time for this interview! The purpose is: ‘help people live their best life’. It is focused on longevity, as people are becoming older and older on average across the globe, with newborns in the Netherlands having a life expectancy of around 100 years, which Aegon does not see as a restriction or annoying challenge, but a wonderful opportunity, as that means everyone’s life is going to look different as they age, including the way they view their own pensions. In the current system, you retire between the ages of 60 and 70, but what will you do with the remaining 30 years of your life if you live to be 100 years old? Longevity, to Aegon, means that we should not think of restrictions, such as if we are going to be able to pay the pensions out, but more along the lines of how we can support people as well as possible in their ageing, such that they can live a long and healthy life, both physically and financially.
This can differ from person to person, some people might want to work until they are eighty, and others wish to scale back when they are 50. This means that there will also be a larger focus on mental health, as with the increasing number of stimuli people receive now, with burnouts proliferating, we see that becoming a larger part of mental health as people age.
How flexible is this purpose?
We are still very much in the beginning phase of this new purpose, and we are working on how to practically implement it in our day-to-day work. The different countries and subsidiaries are going over how this purpose can best be fulfilled, what products are necessitated for this, what this means for the organisations themselves, and, on a more personal level, what does it mean for everyone personally? That is still very much work in progress, where there is a lot of room for innovation within the different country units, as each has their own line of products and culture to deal with, with different implications for this purpose. There has been some progress towards more clearly defining this purpose, as we have identified three behaviours which are part of living your best life. These are ‘we step up’, ‘we tune in’, and ‘we are a force for good’, by which we mean that we want to leave the world a little better than when we found it. There are a number of questions that come from these behaviours, from personal to department level, which we are having constructive conversations about in the office.
This challenge is not one you can easily find at other companies, as it is only by its size and its purpose that you become part of something larger than applying Data Science within Aegon. We are here to serve our clients, and we need to figure out how we can do this through Data Science, and how Data Science can be used to strive towards our purpose.
How does having a centralised data department help with getting the stakeholders on board?
Sometimes, it can be a struggle to get the insights to the stakeholders, but there is always a measure of goodwill. Being a centralised department does help in this, as we come to the customer as consultants and really try to piece apart the problem they are facing. We work for Aegon as a whole, by which discarding a carefully built model would be a waste of everyone’s time. The moment you do an intake, you really try to establish the connection with the stakeholders, and keep an active discussion over what the results will mean. In particular, we try to establish what strategies will need to be implemented given certain results, and to what extent different types of insights and analyses will help them.
What growth possibilities are there within Aegon?
These kinds of possibilities are quite ubiquitous within Aegon. This starts with setting up some growth targets together with your manager, which can take a number of different forms, from specific training courses, to having a chat over coffee with someone, to practising certain skills. Aegon really puts a focus on personal development, which is also incredibly important to me. This was one of my main criteria when searching for a job, as I wanted to join a company that could support my personal development and where I could learn a lot more. The main three questions are: what do I want to learn, what do I need to do to accomplish that, and who can help me with that? I have the fortune that within the department and within Aegon as a whole there is an active discussion on how that personal development goes into play, because this provides the best quality employees for both Aegon and its customers in the end.
The time that these personal development goals can take up differs from person to person and from department to department. It is usually that you are away from your desk for half a day, but you can also embark on longer training courses of a whole week if that is in your development plan. These courses and activities are quite diverse, as they can range from checking another’s code to understand and possibly improve it, to having a speaker or trainer from a third party handle certain topics, such as receiving and giving constructive feedback, personal branding, and other things somewhat more on the soft side of things.
How do you see yourself progressing within Aegon?
I have recently completed the data science development programme, a three year programme that I finished last September, where I am now a fully fledged Data Scientist within Aegon. There are two paths I can take from this point, either I can go to the more technical side through functions such as senior and principal Data Scientist, or I can veer more towards the soft side by becoming either a team lead or a manager. This creates two tracks within Aegon, the more technical and specialised track, and the more management and generalistic oriented track.
Personally, I veer more towards the management track, as I think that maintaining proper communication with the stakeholder’s and coaching my colleagues is a more satisfying challenge than to build a super cool and extremely technically intensive model. In the beginning, I did see it as a bit of a bummer that I was not using all the technical material I learned during my education. However, I have come to enjoy the soft side more over time, which you do not encounter as such during your time in university, so it is something I am looking forward to exploring more.
Did you expect to be more attracted to the soft side of things after your master in econometrics?
I had not really expected it. I had just finished 4 years of study to become as well-versed in building models as possible, both from a theoretical and practical point of view. You want to do complex and difficult things, because that is what you studied so hard for, but then you discover that your clients are not looking for a complex neural network, because they can often be helped a lot more with a relatively simple answer or a different kind of analysis.
What are some examples of projects you typically encounter?
One of the 4 teams in our data science department is quite focused on customer satisfaction, where we have a team which is mainly concerned with customer panels and analyses. A score that we often research is called Netto Promoter Score (NPS), which is a very popular way of measuring customer satisfaction. It is based on a question almost everyone has seen: ‘To what extent would you recommend this product to friends and family?’. That is the question NPS is based on, where it is answered with a score of 0-10. People are divided into three groups based on their score in this question, the first one being the promoters who gave a 9-10 score, the second group being the neutrals giving a 7-8, and the last group being the detractors giving below a 6. The NPS is calculated as the difference between how much percentage of the population is a promoter and how much is a detractor, where a low score indicates low satisfaction and low propagation, while a high score indicates the opposite. This allows us to get an idea of customer satisfaction, which we closely monitor, and always look for ways to increase NPS.
However, there is quite a breadth of processes and data we encounter. For example, another of the 4 teams of our department measures document ‘journeys’ through the organisation, one of which mainly entails following the process of retiring step by step for customers, with the aim of improving both the customer experience and the operational efficiency. I personally enjoy this type of project, as you make a clear map of how the retirement process goes into play, where there are possible bottlenecks, where customers can be served more adequately, which processes are automated and which are not, etc. This allows you to clearly see where there are structural errors in the process, where it takes unnecessarily long, or if you could send the customer a letter at a better time. The question of how to optimise this both for the customer and for Aegon remains an interesting one to tackle at Aegon, where these journeys play an integral part of optimising customer satisfaction.
The length of these projects can differ drastically, from projects which can take just an afternoon, to projects that can take multiple quarters to complete.
The two previously mentioned teams are quite commercially and operationally focused, with the idea of supporting those operations as best as possible. However, this is not all our department does, as we have another team focused on commercial pricing projects, and yet another focused on risk and compliance, something which an insurer such as Aegon is very intimately familiar with.
What projects are you most proud of?
I enjoyed the project in Rio a lot, where we closely worked together with the telemarketing department there. When trying to sell new products to customers, we would usually generate random lists of potential customers, and then start calling, and try to get them interested in buying product B. However, by building a model we could rank these customers by their propensity to buy product B, which allowed us to do some very direct A/B testing, where we would take one call agent who would be given the random list, and another the list generated by the model, and see who sold more at the end of the day. It was really interesting to be involved in the live process of this, where a new generated list meant someone immediately got started on it, which created an incredibly fun game in seeing how the results played out, where we could directly tweak the model as needed and see the effects in action. This direct cooperation gave an immensely satisfying feeling.
Another project which is also of note within Aegon is an investigation into the glass ceiling that women experience. This mainly concerned how the state of affairs was with the promotion of women to the chief executive level, where we analysed what targets we are hitting when it comes to diversity, if we are meeting these targets, where we see possible obstacles, and how the turnover, demotion, and promotion rate differed between men and women, and from which this stemmed.
The discussion of diversity is very active within Aegon, and there has been some measurable success in the Netherlands, especially when you look at the Data Science department. For example, when I look at the chain of command, I personally have a female manager, where we have a female CDO, and a female CEO at the top of the country unit. Across the Netherlands as a whole, the male to female ratio is quite even, but diversity goes further in that and we at Aegon are working hard on creating a more diverse and inclusive workforce. Besides that, we are also pivoting to a more internationally friendly working environment by using English increasingly as the main language of communication within our department, so that international students are also able to join the traineeship and join Aegon. This is very exciting to me, as I would really like to work in a more international department.
What would you recommend to students who want to follow the same path as you did?
My main piece of advice would be, think carefully about what you want to do. Econometrics offers a very large variety of possible career paths, so try different things out. I considered many different things, such as if I wanted to work in consultancy or work as an actuariat, work for a large corporation or a smaller start-up, all of which are completely different working environments and give a very different experience. On top of that, there are a multitude of different sectors you can consider, from finance, to retail, to banking, to investments, to academics. For Data Scientist specifically, job opportunities can be just as ubiquitous, as you can work as a Data Scientist from L’Oreal Paris to Aegon and everything in between. However, when you also consider these different sectors, you could also become a risk manager at ING, a trader at FlowTraders, or a consultant at KPMG. This almost overwhelming amount of possible choices requires that you think very carefully about what you actually want to do and what is important to you. To find this out, grab a cup of coffee with people working in the sector and talk to them, apply for positions at different companies, or reply to LinkedIn messages which you might not have otherwise, and you will get a feeling over time as to what you do and do not like. It can be very hard to find something you want, but finding out what you do not want may be equally as important, as you can then narrow your job search. Try to visit different offices, and ask yourself the question: ‘do I see myself working here 4 to 5 days a week?’. If that is not the case, then try somewhere else.
How long did it take for you?
It took quite a bit of time, which is in part also being a ‘Millenial’, with so many possible career paths, which can make it quite hard to choose. I had the fortune of finding Aegon relatively quickly, but who is to say that your first job will be one you really enjoy, when there are so many other options you can still consider. Furthermore, I encountered Aegon a few times at an experience day which stuck with me, as the job seemed really interesting to me. At a later date, I encountered them again at a networking event, where I really found a good connection with them, so I started to dig deeper to see if I would enjoy the work. I did the same with other companies, and in the end I made the decision to work for Aegon, which I am very content with.
When you look back to the path you walked, what piece of advice would you have appreciated most when you started?
Mainly, I think I would recommend trying different things. When I look back, I am happy that I have been a working student and that I did an internship, as that allowed me to really experience how different companies go to work, through which you get a feeling on how the company operates, how the company culture is, what the office ecosystem consists of, how people interact, and how well these values fit with you. It is mainly trial and error, so try to become a working student, write your thesis at a company, and see if that is what you like to do. This process of discovery through trial and error is extremely valuable in my view.