AI models can adapt to evolving fraud patterns, ensuring better protection without sacrificing user experience
Video games introduce players to exciting worlds composed of stunning visuals, captivating storylines and multiplayer experiences. Strategy, action and community merge on screen as players drive, fly, fight and compete toward the end of their mission.
Behind the scenes though sits another game of sorts, one that is of vital importance for publishers, marketplaces and others in the gaming ecosystem. Storefronts and marketplaces are where players purchase their swords, shields and nitro engines that don't just power themselves up, but the game companies behind them as well. Unfortunately, fraudsters are increasingly lurking right before the finish line, turning seemingly legit revenue into unrecoverable losses.
Fraud in the industry is more than a minor annoyance. In 2023, Roblox reported $110 million in chargebacks. Approximately 10% of all digital gaming purchases are suspected to be fraudulent. To protect against this, most gaming marketplaces and in-game purchases use legacy fraud detection systems that decline anything that looks suspicious. Each year, approximately 25% of all legitimate transactions are declined, which not only cuts into revenue but frustrates and alienates players.
Most gaming companies accept this lost revenue as a cost of doing business. They were able to ride gaming’s incredible growth throughout the 2010s and into the COVID years for heavy profits. However, gaming revenues have flattened over the last few years, leaving game publishers to find new sources of revenue. Approving legitimate transactions that have been declined by existing fraud detection rules would go far in restoring gaming revenue streams. Improving fraud detection accuracy isn't just about reducing losses—it's about turning a business threat into a profitability driver, keeping players happy and ensuring that companies don’t leave money on the table.
Fraud’s toll on gaming is a two-fold problem that impacts bottom-line financial performance and player satisfaction. When legitimate transactions are incorrectly flagged as fraudulent and declined, the gaming company loses out on the revenue. These false positives disproportionately impact new buyers, many of whom don’t return after being inappropriately declined. In addition to the loss of the transaction, the lifetime value of the customer that is lost is significant.
Meanwhile, even long-term players who expect a seamless gaming experience often look for other marketplaces and gaming platforms to spend their money. Churn rates following transaction declines are high and industry estimates report that approximately 40% of those players will move on to new platforms and marketplaces.
False positives occur due to many reasons. Most legacy payment systems use rules-based monitoring to identify fraud. For example, fraudsters often use newly issued credit cards. Many digital goods payment systems will automatically raise a flag when a new card is used for a gaming purchase. The system then recognizes the transaction as high risk and declines the sale.
Rules-based systems do detect fraud, but they are limited in their effectiveness. Legacy systems view each transaction in a vacuum and base their scoring on their ability to authenticate the digital identity of the buyer. Considering that 80% of fraud comes from verified users, though, this methodology is inaccurate and should be retired. Instead, legitimate customers carry the burden of transactional rejections. Players who have taken the time to select skins, items and premium features only to have their choices rejected by a credit card payment system lose trust in the game.
Instead, advanced payment fraud prevention looks at hundreds of thousands of variables to detect even the weakest signals signaling a trend and considers that behavior IS the identity. Each transaction could look legit, but considered together, they are now part of an attack at scale."
Artificial intelligence (AI) is a new player in fraud detection. Rather than following preprogrammed rules to decide on a purchase, AI adapts to detect existing and new fraudulent purchase patterns. AI looks for anomalies within a purchase to determine whether a transaction is fraudulent. Rather than viewing each transaction in a vacuum, it looks at hundreds of thousands of variables to detect the faintest of signals indicating true fraud. On their own, each transaction might appear legitimate, but taken in context with other transaction attempts, it is clear that they are part of a scalable fraud campaign.
For example, AI systems can analyze multiple data sets, such as transaction and keystroke data, to detect fraud. When a user copies and pastes credit card information rather than uses auto-fill combined with a connection from an unusual IP address, it is a strong indicator of fraud.
AI is highly accurate, which leads to fewer false positives and a frictionless purchase experience for legitimate players. AI can reduce the 25% decline rate by 90%, leading to an uplift of 10%-15% in incremental profit.
Through its ability to improve transaction approval rates and minimize fraud, AI transforms fraud prevention from a cost center to a revenue-generating asset. AI-driven fraud prevention also increases customer satisfaction by allowing legitimate transactions to go through more smoothly.
By learning from vast amounts of data, AI models can adapt to evolving fraud patterns, ensuring better protection without sacrificing user experience. This balance between security and convenience helps gaming companies retain more customers, leading to higher transaction volumes and sustained revenue growth over time.
You May Also Like
