Our CTO and Founder is just the right amount of nerdy, lovable, and often times hilarious. Here's how his brain works.
Our CTO and Founder is just the right amount of nerdy, lovable, and often times hilarious. Here's how his brain works.
When I was little, I was always building these model airplanes. For example, there was this Star Wars series of which I had a model kit of…and that was in my bedroom for a really long time.
I also always liked to build huts made from wood. I would build my own secret, multilevel hut that I could play in. You’d get your own nails and wood…so on multiple occasions, I would get a nail in my hand or food because I fell or stood on it.
I also have a scar on my leg because I was drilling a floppy disk drive on my wooden laptop.
Yeah, I built one. But of course, I needed to build a floppy disk into the plank.
But the first real project that I ever built was probably when I went to University. It was a project called IParty that my friends and I started.
Back in the day, you would go to a pub and there would be some people taking pictures of you - ie., during carnivals and fairs. After the party, you could buy these photos for 5 euros apiece but obanly weeks after (of course) because it took time to develop.
So we thought we could make this more fun. We thought we could take pictures, and then party-goers could find themselves online.
Back then, phones didn’t have a camera yet. So the idea was to tag yourself in a picture online and then make a profile from all the photos you went to where your picture was taken. This was before Facebook, in 2001.
We actually pitched it at the university. They hosted a kind of shark tank. But we got completely burned out... Apparently, our business plan sucked. We didn’t have a robust model to make revenue, it was just a fun hobby project.
Exactly! It was really fun, having interactions with people. Because, as a nerd, it’s not easy to make contact with people sometimes. But having this picture-taking mechanism, where people would hang around you, you got a natural interest from strangers. Plus, back then pictures were uncommon. You wouldn’t bring your own camera to a party.
Actually, that got me hired at my internship. I worked at backspace for 7-8 years where I grew from intern to technical lead.
This was in 2003/2004 when I was working on frontend-related technology. One of the founders actually made a web component framework (similar to React, the one we have now, but this was 20 years ago).
It was really far ahead of its time, pushing the capabilities of the browser to its boundaries. Back then you still had internet explorer dominancy. So there were all kinds of weird workarounds we had to get to make it work in a performant and right way.
I joined there as an intern with 4 or 5 people, and when I left the company was 40-man strong. It was quite a journey. I learned a lot - about how browsers worked, how frontend technology works. Which laid the foundation for how different programming techniques work in a browser. I was basically designing patterns applied in browsers.
I switched jobs to JTeam/Trifork from there. After a while, I got approached by someone who worked at Backbase to join a startup Science Rockstars. I was the only tech guy there.
And that’s where I met Leo who joined a little later. It pretty soon became clear that the startup had no future, but working there still afforded me some life lessons.
For instance, how to operationally run a company. To be honest that also sparked the idea of Crobox. When I met Leo we decided to continue with just us two as technical partners in crime. We said goodbye to the old company. Built an MVP version of what we had envisioned in three months. Connected our first customer.
While we were building our MVP we met Rodger through a common friend so we decided to work together for those three months.
Crobox was basically a persuasion profiling tool. Looking back at it, it was really narrow-minded. It focused on conversions, but to provide good insights what you really need are auxiliary services.
For example, we wouldn’t track any pageviews or sessions. We would just track the number of promotions being shown and the effect - e.g., number of clicks or number of add to carts. We could get some numbers but it was hard to convince the customer that the solution actually works.
Now, we can collect insights without needing to serve promotions. Now, we can add or see effects with or without the promotions and gather different perspectives.
The first couple of years were really interesting. My co-founders knew the route to venture capital. I would never find these connections myself IMHO. So we raised 1 million in capital and it became a real company!
When I started Crobox there wasn’t enough money to pay the founders -– yet. So I freelanced about two days a week for some banks. So the three other days a week I could focus on building Crobox.
When we had the VC money it meant I could work on it full-time. Which was quite a revelation, because you get to fully focus on the technology. Then we grew the team to 10-20 people in around 2017. Which was the highest point in hiring people, as well as burning through money.
It’s been a rollercoaster. We’ve had highs and lows. 2017-2018 were called the ‘dark days’, for instance. Many people left. At a certain point, we almost didn’t have any money to pay our bills. So I started to get more involved in the financial part of the company. I’m good with numbers, and I like to match them up so it all makes sense – which is a bit of OCD.
From there we launched the ASICS Shoe Finder in 2018. That made us realize we have a solid platform. It was by adjusting our proposition and finding a better market fit that we could survive. We were 8 people at our low, and we were really a family at that point. Then we found our way up.
From a technical perspective, the technology we’ve built back in 2016 is still proven and still working on our current platform. Our event collection mechanism, which we’ve built up to be scalable, and all the analytics on top of it we basically worked out when we found and built Crobox.
And today they are still performing really well! Of course, they’ve been adjusted and updated to the latest frameworks, but the design patterns are still valid.
For example, we’re using an actor model internally. It’s almost how humans work. I could send you a message, ‘Nikole please get me coffee’. And you either say, ‘sure, I’ll get you a coffee’, or ‘no, I’m busy’.
Of course, I’m waiting for an answer, and the time when you get the coffee is unsure. It’s an asynchronous system of processing messages.
The actor model works like this: You could actually set up a time. If the coffee doesn’t come back in two mins, I’m going to follow up. This mechanism works really well in our internal visitor system. Every message that is for a specific visitor will be sent to that visitor actor.
The actor itself is solely responsible for handling that specific visitor. So it builds up a full context of that visitor inside these actors. It also can respond to questions it gets asked. e.g. Did I visit product X, or see product Y?
Because of this, we can actually do the promotion serving in a very performant way. Everything is processed in real-time and stored in memory. All events related to a product are sent to that specific actor. For example, a question like, ‘how many people bought this product in the last 5 hours’, can be asked to a product actor. Because it knows its own history.
Actor models were used in phone systems back in the day. It can be a low-level mechanism riding in the background, people interacting with our API don’t necessarily know we’re using actor models. But it’s a really scalable solution. For instance, the actors can easily be put into ‘shards’ (breaking it up into multiple things). So we can scale things nicely horizontally.
Building up this system requires you to have monitoring in place. Which is what we learned throughout the past 5-6 years. You build up a suite of different tools that help you achieve your goal.
Sjoerd and Crobox’s Chief of Product, Joris.
It’s in my nature to be dedicated and loyal. I worked 8 years at Backbase, and they had a huge turnover of people. I was kind of their steady force. Which has no correlation…
Jokes aside, I’m invested in Crobox. Both financially and emotionally. I still really enjoy working here, waking up every day, and working on an awesome product.
Plus, the market is so saturated with startups and companies. If I would decide to stop, which one would I choose? I would have to really think hard about that before choosing something else. The thing is, I like to help people. I can set things up and write things quickly, but then people take advantage of these skills. I exploit myself in Crobox to make sure it performs as well as it can.
We try to follow the latest technologies. I’m always on the lookout to see if there are things gaining momentum that we can apply. For example, one of the nicest things we’re using internally is the database called Clickhouse.
We were early adopters, and we have published the open-source Scala driver for Clickhouse we are using internally. We’ve been using it for a couple of years now. I really like how performant and fast it is. It’s one of our current systems that we use for analytics and dashboards.
The founders of Crobox left to right: Rodger, Sjoerd, Leonard.
Currently, we’re hiring for all positions. What we need is a proper DevOps person that can help keep our infrastructure up to date.
And researching what some latest technologies are we can apply so we can keep improving. Also on the frontend, we have some big UI/UX challenges to tackle. And knowledge sharing in the backend, understanding how all our technologies are coupled together, that’s really important.
Finally, extending and improving the product in general by adding new features, additional machine learning, additional reporting mechanisms, and more product finder mechanisms.
The data we provide is one of our USPs. We saw that Google Analytics is really good in communicating web analytics, but what we do is have a lot of many-to-many relationships between promotions, products, and the people that visit them.
Promotions and product relationships allow you to generate unique insights if you correlate these promotions with people’s characteristics.
We searched for solutions out there that provide product analytics and insights…but we couldn’t find any. So by building the analytics yourself, you don’t rely on a product that might break. It’s a really important part of your system that you want to control.
If you use certain third-party systems that are not meant to be used like you intended to, then you hit a barrier on 90% of what you want to achieve. Which is also the question about cloud provider tools – if everyone is using the same what’s your competitive advantage?
If nobody builds an analytical system, and everyone relies on existing analytical systems (google big query for instance), what would your competitive advantage be?
I have two kids, Leah and Isa are 3 and 0. Lea is actually really fun these days. The mind of a 3-year-old is so great, she tried to simplify complex things. For instance, I asked if she had any dinner ‘suggestions’.
She asked if she could have ‘suggestions’ for dinner. Then she asked, ‘what is “suggestions”?’, so I said, ‘it’s like pancakes but different.’
I also like video games. When I have a free night, I like to stroll around in Night City (Cyberpunk). I also played The Witcher.
My partner is my one big (reciprocated) love. We met when we were 19. Yup, she’s my one and only.
We actually met on MSN. Back then, people used to make huge MSN groups with random people. Through a common friend, we somehow ended up as last on one of these random chats. And we talked for a whole year before our first date.
We’ve been together for 18 years now. But that doesn’t seem long. Some people give up fast. But you should find happiness in yourself before finding flaws in someone else and giving up too soon.
Our Employee Spotlights features interviews with Crobox's in-house experts. From data scientists to psychologists, learn what makes us tick professionally (and privately).