Hackathons—intensive, time-limited events where teams collaborate to solve complex challenges—are widely used to build skills, spark innovation, and generate fresh ideas. They create space for rapid learning and experimentation, often leading to practical solutions and new approaches that might not emerge in day-to-day work.

MDRC recently adopted this model to focus on predictive analytics, a data science method that uses historical data to estimate the likelihood of future outcomes for individuals or groups. The goal was twofold: to strengthen staff capabilities and to explore how predictive analytics could inform program improvement and service delivery. The hackathon proved to be an energizing way to refine MDRC’s predictive analytics framework, test new ideas, and build complementary skills like cooperative coding with GitHub.

This post describes what worked well and lessons that may help other organizations use hackathons to drive professional development and organizational innovation. Whether your focus is data science, program design, or operational improvement, the hackathon model offers a flexible approach to tackling complex problems while fostering collaboration and creativity.

Balance flexibility with momentum.

MDRC opted for an extended schedule—eight sessions over four months—to make participation feasible and allow deeper exploration of the topic. This approach worked well for hackathon participants, but it came with trade-offs: Longer timelines require tactics to maintain participant energy and engagement. Shorter, intensive formats can build momentum and focus but may limit participation or depth. The lesson isn’t that one approach is better—it’s that timing shapes the experience. Choose a cadence that aligns with your goals and plan for the challenges that come with it.

Use multidisciplinary teams for deeper insight.

One of the strengths of MDRC’s hackathon was bringing together staff members with complementary expertise—policy specialists who understood program context, data scientists who could translate that context into analysis, and programmers who could write code to train and test models. This mix of people allowed teams to scope projects thoughtfully, choose meaningful outcomes, and interpret results responsibly.

For example, in MDRC’s postsecondary education team, policy experts highlighted that predictions about which students were at risk of delayed degree completion would be useful information for colleges and stressed the importance of incorporating real-time enrollment and academic progress data. Data scientists then worked out how to structure those variables for predictive modeling, while programmers built and tested code to ensure the models ran efficiently. This collaboration meant the final approach was both technically sound and practically relevant, something that wouldn’t have been possible without bringing together people with diverse skill sets.

Practice communicating.

Hackathons aren’t only about building models or solving technical problems; they’re also a chance for participants to practice explaining complex ideas clearly and persuasively. At MDRC, teams presented their progress to each other throughout the hackathon, which sharpened their communication skills and fostered cross-team learning. After the hackathon ended, participants delivered a one-hour webinar with a Q&A to the entire organization, translating technical results into usable information for a broad audience. These presentations required teams to distill complex methods into clear narratives, anticipate questions, and connect their findings to real-world implications—skills that are essential for turning analysis into effects. Creating space for communication—through interim check-ins and final presentations—ensures participants strengthen these capabilities alongside technical ones.

Use hackathons to spark new ideas.

Beyond providing opportunities for staff training and refining MDRC’s predictive analytics framework and code base, the hackathon generated ideas for future applications. Teams explored how predictive information could help colleges better target services, personalize outreach, and identify students or families who need support earlier. These efforts laid the groundwork for deeper collaboration with program partners and opened new possibilities for integrating predictive methods into ongoing work. Hackathons aren’t just about solving today’s problem—they can seed tomorrow’s innovations.

Adapt the hackathon model to different needs.

Hackathons aren’t limited to data science—they’re flexible events that can be tailored to address many challenges. MDRC’s hackathon focused on predictive analytics, but the same structure could further other goals, such as streamlining operational workflows, developing new program models, improving client engagement methods, or exploring qualitative research questions. They can also provide opportunities to build cross-functional skills like facilitating group meetings, developing visualization tools that make group work run more smoothly, and solving problems collaboratively. The versatility of the hackathon model makes it a powerful tool for tackling complex issues in different domains and for different audiences.

Looking Ahead

The hackathon model offers a dynamic and flexible approach to building capabilities and fostering innovation. By bringing together multidisciplinary teams in an immersive, time-bound format, hackathons create a unique space for rapid learning, experimentation, and collaboration. This structure encourages participants to engage deeply with tools and concepts, iterate quickly on ideas, and share diverse perspectives that might not emerge in traditional training or work settings. The adaptability of hackathons makes them well suited for tackling emerging challenges, testing new methodologies, and strengthening skills—ultimately accelerating both individual growth and organizational innovation.

The analyses and conclusions presented herein are the work of the authors listed. Artificial Intelligence (AI) tools were used to help prepare the text.