DraftKings Used ML to Target Promotions at Likely Losers
Internal model identified gamblers prone to losing more when incentivized, raising ethical concerns.
1Algorithmic targeting in gambling
DraftKings developed a machine learning model in 2023 analyzing customer betting patterns to predict who would lose more money when receiving promotions. The system scored users based on their historical behavior, with higher scores indicating greater predicted losses per promotional dollar spent. This allowed precise targeting of free bets and bonuses to the most 'profitable' customers - those likely to lose substantially more than they gained from incentives.
2Ethical concerns surfaced internally
Multiple former employees expressed concerns that the model disproportionately targeted problem gamblers. Data analyst Jayden Butts described the logic as seeking 'good investments' - where problem gambling behaviors became desirable traits. Internal documents revealed tensions between commercial optimization and responsible gambling practices, with some staff questioning the ethical implications of systematically targeting vulnerable users.
3Parallel prevention tool shelved
While refining its loss-prediction model, DraftKings reportedly stalled development of a similar algorithm to identify at-risk gamblers. Four former employees confirmed the company didn't pursue technology that could flag developing gambling problems based on behavioral patterns. This created an asymmetry where systems to maximize losses advanced while protective measures languished, according to the investigation.
4Hundreds of millions in promotions
The model addressed DraftKings' massive promotional spending - hundreds of millions annually on free bets and bonuses - by identifying which offers actually drove incremental losses. Previously, the company lacked visibility into promotion effectiveness. The scoring system provided data-driven targeting, replacing blanket incentives with personalized offers predicted to yield the highest loss ratios per customer.
5Silicon Valley playbook adoption
The investigation frames this as part of gambling operators adopting behavioral targeting techniques perfected by tech companies. Like social media platforms optimizing for engagement, sportsbooks now apply similar data science to maximize 'time spent betting' metrics. This represents an industry-wide shift toward algorithmic optimization of customer lifetime value in gambling contexts.
6Regulatory scrutiny intensifies
The revelations come amid growing examination of sportsbook marketing practices. FanDuel faces congressional pressure over its VIP program, while the MLB players' union advocates banning personalized gambling ads. With DraftKings' revenue already declining, regulators may next examine whether its algorithmic promotion targeting crosses ethical or legal boundaries in exploiting behavioral vulnerabilities.
7Commercial vs. social responsibility
The case highlights the fundamental tension in gambling operators' business models: they profit directly from customer losses. While data science can optimize this profitability, the same techniques could identify and protect at-risk users. DraftKings' choice to develop loss-prediction while deprioritizing harm-prevention tools exemplifies this conflict, raising questions about industry self-regulation.
8Algorithmic accountability frontier
As predictive models increasingly mediate consumer experiences across industries, DraftKings' case establishes important precedent. It demonstrates how even technically sound algorithms can create ethical risks when optimized solely for commercial outcomes without balancing social impacts. The situation may prompt broader discussions about governance frameworks for behavioral targeting systems in regulated industries.
Original article: ReadWrite — DraftKings Used ML to Target Promotions at Likely Losers