What are best practices in data management/reproducibility? The research that I'm involved in has typically used make & git at best, and more typically, hacked

What are best practices in data management/reproducibility? The research that I'm involved in has typically used make & git at best, and more typically, hacked together Matlab scripts.