Embark Studios has finally pulled back the curtain on how they juggle a staggering 100 billion data points generated every single day without breaking a sweat.

During the second day of the NEXUS Co., Ltd. Developer Conference 2026 (NDC26) on the 17th, held at the International Conference Hall in the Gyeonggi Content Agency building, Mathias Andersson from Embark Studios took the stage. The man behind the data for 'THE FINALS' and 'ARC Raiders' explained exactly how his team built a platform capable of tracking every single bullet fired in-game, all while ensuring the player experience remains silky smooth and entirely unaffected by the heavy lifting.

According to Andersson, the sheer volume of data is mind-boggling. They track over 1,000 different event types, hitting peaks of 100 billion events per day—that is a massive 30 terabytes of data daily. But it is not just about hoarding digital logs; they treat this data as a vital asset to keep the games fair, fun, and competitive.

The primary goal, as you might expect, is security. By tracking every trajectory, they can instantly spot shots that defy physics or movement patterns that are simply impossible for human players. Beyond catching bad actors, they use the data to keep the game balanced—tweaking weapon damage or accuracy if something feels a bit too 'God tier game'. They even track who fires first in a skirmish to gauge player style, helping to match aggressive players with like-minded opponents.

Now, you might be thinking: collecting hundreds of billions of events in real-time surely drags the game down or costs a fortune? Andersson insists that while the game itself is complex, the data aggregation should be anything but. He noted that they chose a 'strictly isolated architecture' to ensure that the data collection pipeline never interferes with the game’s actual performance.

The team essentially parked a custom application right alongside their Unreal-based game server. The server bundles the barrage of firing events into small, 30-per-second batch processes before passing them off. This keeps memory usage focused on tracking real-time state changes, such as player movement and cumulative damage, without the overhead.

Once a round wraps up, the data is tidied into a single file, negating the need for a bloated, complex database. It is a delightfully lightweight and intuitive approach that turns raw logs into concise summary statistics efficiently. Because they designed this bespoke bundling system themselves, they manage to keep costs surprisingly low even when dealing with massive datasets.

Andersson also hammered home a vital point: the data team should never be the sole gatekeepers of information. He advocates for 'Democratizing Data', ensuring that everyone—from developers to map designers—can poke around the data themselves. Their star tool? The 'Round Viewer'. It is a simple, no-nonsense list page that lets anyone pull up player lists, server error logs, or a specific player's pathing for the day. It has quickly become the most versatile tool in their arsenal.

The 'Map Replay' tool is another stroke of genius, visually reconstructing how players moved through a round and where they took their final shots. It is an invaluable way for map designers to spot geometry exploits or suss out cheaters moving at unnatural speeds. They have even developed an Unreal engine plugin that overlays heat maps directly into the editor, allowing them to see player interactions in real-time without ever leaving their workspace.

In short, if you want to process massive real-time game data without burning a hole in your pocket, you need a strategy rooted in simplicity and efficiency. But perhaps more importantly, you need a vision that prioritizes democratizing that data, making it a shared asset for everyone building the game.