Study: Does Access to Human Coaches Lead to More Weight Loss Than With AI Coaches Alone?
Adding human coaches to artificial intelligence-powered weight-loss programs significantly boosts user success, underscoring the value of hybrid human-AI models in digital health, a new study suggests.
Drawing on data from about 65,000 users of the HealthifyMe app over three years, the researchers found that individuals using both human and AI coaches lost 74 percent more weight over three months than those relying on AI alone. The work highlights how human support can enhance motivation and accountability in tech-driven behavior change.
"While AI offers scalability and efficiency, our findings show that the human touch still plays a critical role in achieving meaningful outcomes," said Puneet Manchanda, professor of marketing at the University of Michigan's Ross School of Business.
The study examined users who subscribed to two types of paid plans on HealthifyMe, a health-and-fitness app founded in India. The less expensive option, costing about $44 per month, provided access only to an AI coach called "Ria." The premium plan, priced at $50 per month, included both the AI coach and access to human coaches, who worked with users to develop personalized nutrition and fitness plans, offer encouragement, and answer questions throughout the weight-loss journey.
To ensure a fair comparison, Manchanda and colleagues Anuj Kapoor of the University of Missouri and Sridhar Narayanan of Stanford University used statistical matching techniques to account for factors such as age, gender, starting weight and app usage patterns, minimizing potential bias between the two groups.
The study found, on average, that users with access to both AI and human coaches lost 2.12 kilograms (4.67 pounds) in three months, compared to 1.22 kilograms (2.69 pounds) for those using only AI coaching.
The presence of a human coach also changed how users interacted with the app. Those with human support:
- Set more ambitious weight loss targets (17.6 kg vs. 15.2 kg)
- Logged meals nearly twice as often (102 vs. 49.7 times per week)
- Monitored their weight more frequently (1.35 vs. 1.12 times per week)
These differences suggest that human coaches helped users stay more accountable and committed.
"Human coaches appear to increase the intensity of engagement, leading to more consistent behaviors and better outcomes," Manchanda said.
The benefits of human coaching varied across different user groups. Women showed greater relative improvements than men, and older users benefited more than younger ones. Surprisingly, users with a lower starting Body Mass Index saw greater benefits than those who were more overweight.
The researchers reasoned that while users with higher BMIs might benefit from the empathy and customization human coaches provide, they may also experience greater embarrassment when interacting with human coaches, potentially offsetting some advantages.
The study's findings carry important implications for consumers, companies and policymakers alike.
For individuals considering fitness apps, the research suggests that premium plans with human coaching may justify their higher costs, particularly for women, older users and those with lower starting BMIs. However, the researchers also noted that while the 74% improvement in weight loss outcomes sounds significant, the absolute difference is modest-about 2 pounds over three months compared to 2.7 pounds. Consumers should manage their expectations and weigh these realistic benefits against both the financial premium and time commitment required for human interactions, they said.
For health tech companies, the study offers the first large-scale evidence that human coaches provide measurable value beyond AI tools. Given that coaching is costly and hard to scale, companies can use these insights to strategically allocate human resources to the users most likely to benefit, optimizing both outcomes and operational efficiency.
From a policy perspective, the research underscores the importance of hybrid solutions in combating obesity-one of the world's leading health challenges. While AI holds promise for scalable, cost-effective interventions, the study suggests that fully automated tools may fall short for certain groups. Tailored strategies that combine AI with targeted human support could deliver more equitable and effective results.
