THE PRODUCT PROBLEM
In the realm of assistive technology, the challenge was to create a non-invasive method for individuals with brain lesions to communicate effectively, without the need for surgical interventions. Existing non-invasive solutions were plagued by low accuracy, rendering them impractical for real-world use. The competitive landscape was dominated by invasive techniques that, while accurate, were not scalable due to the inherent risks and costs of surgery. Meta needed a breakthrough to bridge this gap, providing a viable alternative that could be adopted widely and improve the quality of life for millions.
THE DECISION
The pivotal decision was to prioritize data scaling over architectural innovation, aligning PM, engineering, and research teams around a shared vision of leveraging massive datasets to enhance model performance. This decision required a commitment to open collaboration, as evidenced by the release of both the model and training data to the public. The alignment was challenging, as it involved balancing the need for rapid improvements with the long-term goal of fostering community-driven advancements. In hindsight, this tradeoff proved advantageous, as the significant performance gains were achieved primarily through increased data volume, validating the decision to focus on data rather than immediate architectural changes.
THE LESSON
This case underscores the critical importance of data in driving product innovation, a nuance often overlooked in PM/engineering collaboration. While PMs may traditionally focus on feature sets and user interfaces, this project highlights how data accessibility and volume can be the real catalysts for breakthrough performance. The collaboration between Meta and the wider research community illustrates the power of open-source initiatives in accelerating progress. PMs should consider how fostering open ecosystems and prioritizing data scalability can unlock new levels of product capability, even when technological breakthroughs seem out of reach.