The useful thing about a failed game is that players can still tell you what they wanted. Bagelcode’s first serious attempt was a real-time puzzle title for Kakao. The engineering was difficult. The business model was not difficult enough: according to co-founder Ilhwan Yoon, once somebody had spent about ₩50,000, there was little left to buy. The game went nowhere. Inside its shop, however, sat a simple random slot machine. Players kept spending on it.
That tiny wheel was not the business, but it was a clue. Yoon said the team realized games were not only about control or competition; randomness could be its own source of anticipation. Five developers who had already tried webtoons, a fashion blog, content recommendations and social dining turned toward social casino - a category none of them knew well, aimed at American consumers they did not resemble.
01 / The clue inside the miss
They picked a market from the chart, not the mirror
The founders studied the 150 top-grossing apps on Google Play in North America. One hundred were games; roughly a quarter of those were social casino. It was an almost aggressively unromantic act of market selection. Instead of asking what five Korean developers felt inspired to make, they asked where consumers had already demonstrated a willingness to return and spend.
The missing ingredient was cultural knowledge. Bagelcode made a list of about 50 casino-game companies and called them. Yoon, then on his honeymoon, used a London stopover to meet prospects. The resulting network put planners in Denver, creative talent in San Francisco and London, and the core development team in Seoul. Remote collaboration arrived at Bagelcode before remote work had a wardrobe.
The side feature was the market raising its hand. Bagelcode was willing to notice.What changed the founders’ mind
Vegas Party Slots became the first proof. Released in 2016 after a publishing agreement with Big Fish Games, it accumulated more than three million downloads and ₩10 billion in sales by the time the founders spoke publicly in 2017. That success attracted capital. But it also sharpened a problem: Bagelcode did not fully control the route from its studio to the player.
02 / Distribution is a product
Two deals vanished. The lesson stayed.
First, a Korean publisher backed away after its own game became a hit. Then a US publisher was sold, and Bagelcode’s agreement disappeared with the transaction. Neither event was a verdict on the game. That was precisely the trouble. The founders decided they could not leave the company’s fate inside someone else’s priorities, cap table or filing cabinet.
Club Vegas, globally released in 2017, was built and published directly. It offered the familiar slots fantasy but added clubs, events and a long calendar of fresh content. The customer is not a casino operator. It is a consumer looking for a light, social, repeatable diversion. Bagelcode’s terms are explicit: these are free-to-play entertainment products, not real-money gambling, and they do not award cash prizes. Revenue comes mainly from optional virtual currency and goods, with advertising in some products.
Owning distribution did not mean doing everything alone. In 2018 Bagelcode paid £18.1 million for JPJ Group’s social business, adding a London operation, experienced people and titles including Jackpotjoy’s social slots lineage and Bingo Lane. This was capability by acquisition: a Korean startup buying years of category memory in the market it wanted to serve.
03 / The compounding machine
Twelve terabytes a day, one boring improvement at a time
Bagelcode’s difference from other mobile studios is not that it possesses data. Every live game emits exhaust. The difference is organizational: product, user acquisition and finance are expected to read from the same dashboard. A 2023 interview with the company’s Data & AI director described more than 12 terabytes of new data arriving daily. Teams automated parts of A/B testing and used models to estimate when advertising spend would be recovered.
This matters because free-to-play math is delayed. An ad costs money now; a player may return value over weeks or months. If a studio cannot connect the campaign, the player and the eventual purchase, it can buy impressive download charts and an ugly bank account. Bagelcode’s loop compares variants, watches retention and revenue, predicts payback, then shifts budget. The team has said it will choose a version that is only 0.1 percent better. There is no cinematic montage for that.
Profit finally joins the party
Reported annual revenueKRW billions
The method compounds across titles. Lessons from Club Vegas informed Cash Billionaire; experiments could travel in the other direction too. Performance partners supplied more evidence. A Windows campaign produced reported return on ad spend 150 percent above iOS. A connected-TV test began with a modest budget divided across audiences, paused weak groups, adjusted frequency and time slots, then expanded to more than twice the initial spend. The pattern is consistent: earn the right to scale.
04 / What to steal
Copy the loop, not the slot machine
Study spend, not applause
Use rankings, transactions and retention to choose a market. Compliments are pleasant; repeated behavior is a better compass.
Keep the odd clues
Instrument side features. The unexpected pocket of activity may reveal a stronger job than the product was designed to do.
Join product to distribution
Give builders visibility into acquisition cost and give marketers visibility into retention. Otherwise each team optimizes a different company.
Scale after proof
Cap the first test, define a payback threshold, remove weak segments and increase spend only as measured performance survives.
The company’s cost was not a single launch budget. Publicly visible commitments include four major funding rounds through Series D, totaling ₩77.3 billion by Bagelcode’s count, and the £18.1 million UK acquisition. The harder cost is time. The company was founded in 2012 and did not report its first annual operating profit until 2022. A decade is a long runway for a lesson about patience.
By 2024, Bagelcode reported ₩111.7 billion in revenue and ₩13.8 billion in operating profit. Club Vegas had crossed $250 million in cumulative sales. Cash Billionaire was growing too. Those numbers place Bagelcode among meaningful global social-casino publishers, though it still competes with larger operators such as Playtika, SciPlay, Product Madness, DoubleDown and Huuuge Games - companies with deep catalogs and formidable acquisition budgets.
When the playbook breaks
This approach is weak when traffic is too small for reliable experiments, creative production cannot feed frequent tests, privacy changes damage attribution, regulation constrains the category, or the business cannot finance a delayed payback. Data can rank choices; it cannot manufacture taste. A team that optimizes only what it can count may polish a game nobody loves.
05 / The genre escape
Now Bagelcode has to prove the system travels
Social casino remains the engine, but Bagelcode is trying not to let a successful category become a comfortable cage. Its current catalog includes Golden Mango Casino and Bingo Lane alongside Pack & Clash, a backpack-based strategy roguelike; Town Fall, a tower-defense RPG; and Blood Invasion, a darkly comic idle-action game about a vampire defending her castle from humans.
Blood Invasion is the cleanest test. Bagelcode developed it and SayGames published it globally in July 2025 after several months of soft-launch iteration on play, systems and monetization. SayGames reported more than 250,000 downloads during the global launch weekend. It is too early to place that beside Club Vegas’s long sales history, but the arrangement is revealing. After learning not to surrender its fate to a publisher, Bagelcode did not conclude that partners were bad. It learned to be precise about what each party contributes.
The same pragmatism now shapes its AI push. Bagelcode says employees use role-specific agents and tools across engineering, data, art and support. In 2026, executives described mornings when developers review pull requests prepared by agents overnight. The claim is eye-catching; the durable question is familiar. Does the new workflow improve the next decision, and can the result be measured without confusing motion for progress?
That is where Bagelcode fits in the market: part game studio, part performance publisher, part distributed operating lab. Its customers see reels, clubs, vampires and cheerful treasure. Behind the glass is a more sober machine - one built after failure taught the founders to look closely, and broken deals taught them which levers they needed to hold.
The best Bagelcode lesson is not to bet on randomness. It is to build a company that gets less random with every release.The operator’s takeaway