Inside BMG – tech talk with Sebastian Hentzschel

Sebastian is Group CTO and EVP of Royalty Processing Shared Services for BMG. The Bertelsmann’s music publishing and recorded music division is headquartered in Berlin and operates in all major music markets in the world. In 2018 it turned around 545 Million Euro according to Wikipedia.

I met Sebastian not so long ago and we immediately had a click – same old, same old – technology, startups, company culture. I thought would be cool to get a sneak peak at BMG’s tech stack and here is the result – my first interview of this kind.

Sebastian, what is your technology stack?

Our primary stack is AngularJS, Java Spring, Elastic, SQL Server. The other two stacks we have is .NET for one of our key data pipelines on top of MSBI, and Hadoop/Spark for our royalty engine and data analytics platform.

What products do you work on?

We’re currently actively working on roughly 15 products. The most prominent is MyBMG, a web and mobile app our clients can use to see their realtime income earned globally from their music and can run data analytics on it. Other products include BMG’s global repertoire licensing platform our teams worldwide use for finding, pitching, quoting and licensing our 2m+ music rights to business clients in the film, TV, advertising and gaming industry. Lastly, we’re currently expanding our rights and royalties platform across all business segments, moving from traditional to big data technology for better scalability.

Which one are you most proud of? Why?

I’m proud of us as a company. We’ve built a business from the ground up, starting in 2008 with minimal revenue and three people in Berlin and growing to more than half a billion in revenues and 800+ people worldwide.

From a tech perspective, the business of music and music rights management is a very data intensive business. We started out with basic off the shelf software and quickly had to build a new technology platform from the ground up while not risking business continuity. We’ve hit new, growth driven inflection points almost every year, and we’ve been able to get through them and come out better on the other end. Also, we’ve integrated acquired companies globally very tightly on to our platform, which is something that wasn’t easy but we’re clearly benefiting from today.

Just curious – are you a Mac or Windows person?

At home I use Mac and Chrome OS. At work I need my Windows, mainly for its keyboard shortcuts.

Which apps you use daily?

Let me check … FT, Koyfin.com, Whatsapp, Evernote, Headspace, Teams, Jira, Confluence, Trello…  well, and Outlook & Excel, of course.

Let’s go back to tech. What is your development process like?

Our application management and product management teams are internal, software engineers are outsourced with a handful of very close development partners of ours. Most of our products are developed in agile mode. As a result of outsourcing, we have a proxy product owner on the vendor side who takes on the agile process ownership and closely collaborates with our internal product owner. Agile doesn’t fit for all projects, however. We run our royalty platform development project in waterfall with monthly releases. For us, this works better for the type of projects with zero acceptable margin of error.

Tell me about your team. What does your team structure look like?

Teams are organized under our VP of applications & infrastructure, VP of technology product, head of data analytics, as well as three regional heads of workplace services for the US/Latam, UK and Europe & Australia. Software engineers, data center infrastructure management, and data engineers are all outsourced to selected vendors, who we work together very closely. We intend to keep the relationships with those vendors consistent so as to leverage learning curves, knowledge and efficiencies. But we do retain a healthy dose of competition between them.

What is your preferred development mix (e.g. in-house, nearshore or offshore)?

Again it really depends. We operate a mix of onsite and offshore. We chose to outsource development primarily for reasons of scalability. We’ve been needing to spin up, or ramp down, new teams rather swiftly. For example, when we started to replatform our royalty engine, we needed a big data team quickly. Our partner was able to put together that team within a month. What works best though is when developers or business analysts are co-located. On the flip side, the higher the time zone difference between your product/business and development teams, the more difficult collaboration gets. Given that offshoring usually involves outsourcing to companies that have their cultural roots outside of European or Anglo-American culture, projects success requires people who understand both cultures very well and can translate between them.

What I found important to look for in a vendor is that they bring their own set of processes, best practices and innovation to the table. That’s a huge value added.

How do your scale your ops when you have to?

With the growth we’ve had as a company, our capacity planning has frequently failed us with data volumes outpacing quickly our “worst case” projections. On the flip side, the nature of our business cycles is rather solid and doesn’t come with massive spikes of usage. This is why we run data-crunching intensive processes in a private-cloud data center on “our own” virtual and physical servers managed by a third party company. This allows us to significantly leverage fix cost/capex investment.

How do you hire? Do you do brainteasers or coding challenges?

I’m trying to maintain a comfortable atmosphere so the candidate has enough space to present him or herself. I may chance pace should I notice the candidate is vague or evasive. Yes, I do case studies and brainteasers. I want to understand how a candidate thinks, the cognitive avenues they’re taking and how they manage challenges in a high-stake situation. I’ve had candidates really stepping up in these situations, or falter, not because they couldn’t find the answer right away but they wouldn’t engage fully.

What makes a candidate stand out?

For the roles we hire for, I am looking for technical experience and intuition, as well as empathy and the ability to make human connections. In product management, the ability to both be empathetic and structured is important for me. In the application dev and ops space I look for a sense of relentlessness when it comes to driving constant improvement. In any case, I look for authenticity and the candidates who stand out are those who are strong in what they know and do, but are humble at the same time.

Let’s talk architecture. How do you see your architecture in 2-3 years?

I would love to move to serverless architecture, although this may still take awhile. Most likely, I’ll be pushing more services to public clouds where the re-architecture and operational model is financially sensible.

Do you measure speed and quality of delivery, and how?

What’s most important to me is that the entire team – business, business owner, product owner, proxy PO and developers – find their own groove and iterate on that. That’s the most sustainable and ultimately fastest model.

If a feature can be implemented either quick and dirty, or slow and perfect, what would you pick?

It depends. For client royalty obligation calculation it’s clearly slow and perfect. We’re trying to use the concept of going live quickly with a minimal viable product a lot though. It does however depend on the readiness and willingness of the respective business function to engage with that concept.

Do you have dashboard and what metrics do you track?

Yes. My weekly capex investment and operational spend. There’s no week that goes by without me looking at it. The non-financial figures I look at are our global incident burndown as an indicator for technology health and a supply and demand balance. I’ll be introducing a weekly team barometer shortly to ensure team health for both projects and operations – previously our team was much smaller and I was able to speak with almost everyone regularly. This has changed.

What mega trends already influence your products? What is coming next?

A few things. A focus on holistic customer experience is something that influences more and more of our products, even internal ones. Big data technology is no longer a strategic vision but operational necessity. My approach is to utilize all data available and automate reporting and data analysis. We’ve also started integrating machine learning in the first product of ours. Not all problems are machine-learning problems but we have identified a few areas in front and back office that would benefit from this technology. Blockchain is something we’re looking into and experimenting with but given what we know it’s a technology that may not have the impact it has long been proclaimed it would have.

Any question I should ask you?

When is our next lunch?

Sure thing, it was a pleasure meeting you and hearing your thoughts. Looking forward to see you soon.