Playing without data is like a cake without walnuts – the power of data analytics in gaming

Nikola Vasiljević,
Analytics & Insights Manager, Mad Head Games
There aren't many experts in data analysis in gaming, and those who are involved in it often build something from scratch. That's why Nikola Vasiljević (Analytics & Insights Manager, Mad Head Games) decided to establish a peer group with colleagues from the industry to learn and build together this small but important part of the gaming community.

SGA: How would you describe data in gaming? What types of data do we measure, and what insights do we extract from that data? Which departments within a company can/should rely on data? What tools are used?

Nikola: We strive to look at our players holistically and want to know a) what they do in games and b) what their experiential journey is like. We measure this through two sources of data – telemetry analytics, where we record player behavior (jumped, shot, died…), and research, where we talk to players and try to learn more about their experience; we discuss what they liked, what didn’t, what they would change, what was difficult or easy, starting from general questions to very specific and targeted ones related to different aspects of the game – combat, animation, sound, atmosphere… And then even more granular 😊 It’s crucial to keep asking the question “Why” until you reach an axiomatic answer and a real understanding of the player’s actions and experiences.

The goal of our team is to have a data-driven organization. When it comes to game design, we primarily collaborate with game designers, but also with the leads of other sectors (sound, programming, level design…), and this collaboration is becoming broader. As we actively educate more colleagues, more colleagues find ways for us to help them. An interesting example is a recent collaboration with the narrative design team. Who would have thought that “Excel wizards” and creative narrative designers could find common ground in analytics? 😊 Besides games, we engage in analytics in all fields – we’ve created tools for HR and PM teams. Where there is data, we are there too.

For these purposes, we use various tools. Part of our constant stack includes Tableau, MySQL, MongoDB, BigQuery, and Jupyter Notebooks (Python), but there are many more. Of course, Excel must not be forgotten, nor PowerPoint 😊

SGA: What does the data team do at MHG?

Nikola: Starting from a high level, our mission is to create a data-driven organization and to support all decision-makers with findings and recommendations that will help them better understand the issues and make more educated decisions.

If we talk only about supporting game design, we have devised and implemented our framework that supports the development of a video game from the idea to post-launch. Within that framework, we:

  • Research market conditions and trends
  • Player segmentation
  • Analysis of the sales potential of the proposed game through sales forecasting
  • Competitor analysis
  • Analysis of the preferences of the target audience to find out what attracts them to a certain genre
  • Assist in defining the pillars of our game
  • Create a plan for analytics and research
  • Create a telemetry system
  • Create machine learning-based analyses (we perform sentiment and topic analyses of player feedback left on the internet)
  • Conduct playtests
  • Analyze data from UX research, telemetry
  • Communicate key findings and recommendations
  • Participate in defining action steps
  • Conduct follow-up tests and post-mortem analyses

Or, in simple terms, we analyze all available data to help our team make better games in line with player expectations.

SGA: What is your role specifically?

Nikola: Initially, it was to conceive what our sector should do, as I was its founder. Now, I would say my role is to understand the needs of our business and based on that, devise a strategy for our sector and ways of implementing it, ensuring that we are innovative both in the use of new technologies and methods and in our approach. In practice, I lead the analytics and research team and coordinate the sector’s work, including colleagues from other sectors (shoutout to our technology and programming teams and our “Tuesday syncs” without whom we wouldn’t survive 😊), with the ultimate goal of ensuring that every decision-maker receives timely insight based on which they can make a decision and determine actionable steps. Part of my duties is, therefore, managerial, but I am still very involved in operational work. It’s something that fulfills me, and I don’t relinquish it 😊 However, I strive for my main contribution to be at a strategic level. To exist as a business link, to ensure that everything we do has practical applications, to devise new initiatives and improvements. Because there is always room for improvement. And, of course, to pass on my knowledge. I am very proud of everything we have achieved so far, but at the same time, sometimes I am intimidated by all that is on my mind that we haven’t done yet because the potential for using data is truly enormous, and it’s not easy to keep up with the dynamics of everything that is happening. But, we take it step by step 😊

SGA: Obviously, you care a lot about what you do, and that’s how the idea for this initiative came about? Does it have an official name? What are its goals?

Nikola: Now it can be said that the official name is Data in Gaming Meetup. The idea arose after last year’s Data Science Conference, or Pandora event, which brought together gaming data professionals. I had the opportunity to give a lecture on “Boosting Game Design with Analytics,” and the reaction from industry colleagues was great. We met, exchanged experiences, and it was generally evident that there was a healthy energy and potential for something more. And it was simply a shame to wait another year for someone else to bring us together. So, I decided to take matters into my own hands. The idea simmered for a while, and I thought about what to do with it. The idea is great, but the realization of additional activities can be a commitment, and that can be discouraging. However, with the support of colleagues from MHG, especially the communication team, we easily and efficiently put it into action. Of course, the precondition was the support and response from colleagues in the industry. So, last summer, we met at a casual gathering in our studio, where I presented the idea, and the reaction was positive. I didn’t expect anything else, or I wouldn’t bother with all this. Some really great people work in our industry.

The meetup format is such that we have a formal, educational part where some methodology, business case, interesting topic, anything from the field of analytics and research is presented. The second part is, of course, relaxed; we socialize, chat, eat. So, the goal is, on the one hand, to exchange experiences and learn from each other, and on the other, to connect and create a community.

SGA: You’ve had two meetups so far to test the waters. What was the situation like?

Nikola: As I mentioned, the initial reaction at the first meetup was positive, and somehow everything settled even more at the second one. People’s retention was decent 😊 Everyone became more relaxed, got to know each other, chatted. The formal part was useful… I hope… I mean, there were a lot of questions, that must say something 😊 Several colleagues have already expressed a desire to host the next time or to present at our place. So, I would say there is potential for our caravan 😊 That’s why it’s essential to support it from the start, provide an influx of new energy until everything gets rolling.

...And in the last year, the topic of sentiment analysis has become a real hot topic, along with the development of Natural Language Processing (NLP) and Large Language models, such as OpenAI's... I will be talking about how we have integrated all of that and the results we have achieved, along with practical advice on how it can be done at Pandora.

SGA: What will all interested individuals have the opportunity to hear at future gatherings? What formats/speakers/topics do you have in mind?

Nikola: At the first official meetup, we presented our 360 framework for analytics and research, through the theme “THE STRATEGIC ROLE OF ANALYTICS AND RESEARCH IN VIDEO GAME DESIGN.” There were other candidates for a topic, from emotion analysis in reviews to data pipelines, data protection… The spectrum is wide. The idea is that from the next meetup, another studio shares its experiences, and maybe even hosts. We will discuss topics on the fly. We don’t want to create any additional pressure or obligation from this initiative but rather have those who really want to participate and enjoy doing so. And I am sure that everyone who is already part of the initiative has a lot to share. Just in the casual part of the gathering, I heard so many interesting topics. You know, having a conversation with a few colleagues on a very interesting topic, such as intelligent dynamic systems for AI NPCs or speedrunning Super Mario, and at the same time, overhearing what another group is talking about, and that is ultra-interesting. There is enough material, that’s for sure! The idea for the future is to slowly expand the story, not only among analysts and researchers in gaming but also among all those for whom it can be interesting and useful. And for those it is, there is also a form for registration 😊

SGA: Just to clarify – is all of this for both the core and mobile parts of the industry?

Nikola: Absolutely.

SGA: Will you also be giving a lecture at this year’s Pandora conference? What will you be talking about?

Nikola: That’s right. My lecture concludes the Pandora program on Wednesday, November 22. The topic is “Leveraging Sentiment And Topic Detection For Decoding Player Feedback.” In the last year, we have been very intensively analyzing comments and reviews that players leave on Steam, YouTube, Reddit… We were interested in what they were talking about and what their stance was. We did analyses for the game Scars Above, which we released in February, but also for competitive games, and we built a new tool that became part of our stack. Simply put, players give you all that free feedback; such an opportunity should not be missed. And in the last year, the topic of sentiment analysis has become a real hot topic, along with the development of Natural Language Processing (NLP) and Large Language models, such as OpenAI’s… I will be talking about how we have integrated all of that and the results we have achieved, along with practical advice on how it can be done, at Pandora.

SGA: Which other lecture would you highlight from Pandora?

Nikola: Oh, this is a really tough question. I’ve been planning my visit to lectures and online sessions for days, and my head hurts because I don’t know how to fit everything in 😊 It seems that there is much more material than last year, so I’m really in a sweet dilemma. If we look only at Pandora, it seems much more diverse than last year. It depends on what interests you the most, data protection, analytics, research. If I were to highlight one lecture, let it be the introductory one – Konrad Tolmar, Research Director from EA, mainly because I think this is a unique opportunity to hear something like that. But if you are really interested in gaming, go to all Pandora lectures. I did that last year, and I didn’t regret it.