How lean multiplayer game studios are outshipping teams ten times their size

Why retention economics matter more than launch numbers

The studios winning long-term are building products people return to. The numbers on this are stark. Median Day 1 retention across mobile games sits at 23%. By Day 7, you're down to 7–8%. By Day 30, under 3% of players are still around.

Research from Bain & Company puts the financial implication clearly: a 5% improvement in retention can increase profit anywhere from 25% to 95%. The games that cracked this are showing what's possible:

These are business outcomes that come directly from treating live operations as a first-class part of the product from day one.

How multiplayer drives player retention

The data on co-op and multiplayer is hard to ignore. R.E.P.O sold 13–17 million copies. Peak hit 10 million copies and was built by a seven-person team. Schedule I peaked at 459,000 concurrent players. These are not massive studios with decade-long development cycles.

Cross-platform play matters. Games that support it see 31% more daily returns and 45% higher engagement in the first 30 days. Players who can jump between mobile, PC, and console without losing progress or friends come back more often and stay longer.

Photon handles 1.5 billion monthly active players across its network. Stumble Guys runs on Photon at 1 million concurrent users, generating $2M in daily revenue at peak. The infrastructure for real-time multiplayer at this scale is solved. Your team should be building game.

Build vs buy: the case against a custom game backend

Backend engineering for a live service game is deep, specialized work: server authority for cheat prevention, matchmaking, game config management, economy balancing, liveops tooling, analytics pipelines. Doing it yourself means hiring people who won't be working on your actual game, and accepting the ongoing maintenance burden forever.

The whitepaper follows two studios building real-time multiplayer games right now. Different genres, different cities – same bet on infrastructure over headcount.

Future Run is an eight-person team in Tampere, Finland, building Riot Ball – a real-time multiplayer sports game. They're using Photon Quantum for networking and Metaplay for backend. Their CEO, Riku Rakkola, put it plainly: "We don't have a DevOps person." They've had zero backend-related problems at scale. Eight people, real-time multiplayer, no backend team.

Antihero Studios is twelve people in Barcelona, working on Misfitz – a 20-player multiplayer action game. The team includes veterans from Supercell, King, and Bandai Namco. CTO Andre Parodi had a clear position from the start: "I've thought very strongly that we shouldn't be developing tech." They evaluated Metaplay with a two-week test branch. A year later: "Absolutely zero regrets."

Both studios are competing with the output of much larger teams. That's a direct consequence of the infrastructure choices they made early.

How Metaplay and Photon split the multiplayer stack

The division of responsibility between the two affects how you'd actually structure your architecture.

Photon owns the real-time layer: networking, matchmaking, and the scalable infrastructure underneath it. Photon Quantum is their framework for deterministic simulation – useful when you need lockstep multiplayer physics, like Riot Ball. Their CTO, Christof Wegmann, has identified the failure modes he sees kill games at scale: session drops, matchmaking that can't fill at low player counts, debug logging left on in production. Photon's Quadrant framework gives teams a way to choose the right networking architecture for their specific game type before they build anything.

Metaplay owns the live-service layer: progression systems, economy, game configuration, and live operations tooling. Every studio gets full source code access from day one – no dependency on us to make changes, no black box to work around. Server-authoritative design is the default, which matters for cheat resistance in competitive multiplayer. The studios already on Metaplay include Supercell, Metacore, Trailmix, and Lessmore. Cost runs at roughly half what comparable alternatives charge.

Our CPO, Teemu Haila, has a few lessons from those deployments that come up consistently: metagame matters earlier than most teams plan for; you won't know exactly what backend features you need until you need them, which is why source code access matters; and server authority is foundational if you care about fairness.

Where the multiplayer games market is heading

Geography is part of the story. Southeast Asia generated $6.2B in 2025. India is projected to triple to $7.7B by FY2030. These markets are mobile-first and social – the kind of games where multiplayer backend infrastructure and liveops tooling are table stakes.

The consolidation trend – fewer titles, more playtime concentrated at the top – pushes in the same direction. Building for retention from day one is the baseline. A single launch window doesn't sustain a studio.

The whitepaper has the full data and both case studies.