Industrial policies combine multiple instruments, yet existing research typically studies individual components or treats the bundle as a monolithic intervention. I examine complementarity in policy components of an agro-processing industrial policy, Operation Flood, the world’s largest dairy-development program implemented across Indian districts from 1970 to 1996. Digitizing the implementing agency’s annual reports, I construct a district-by-year map of farmer cooperatives and dairy-processing plants. The components’ staggered rollout generated some districts with one component and some with the bundle. Using matched stacked difference-in-differences, I find that village milk prices in bundle districts increase by 33% during the flush season. Neither arm alone shows any significant effects, and I reject additivity for prices at the 10% level. A stylized model explains this complementarity. Cooperatives alone can saturate local markets, while plants alone cannot transmit the value of an additional outlet upstream. The bundle routes surplus milk to processing and passes value back to farmers. Turning to structural change, the industrial employment share increases by 1.7 percentage points in bundle districts, with no significant standalone effects. This increase comes mainly from non-food manufacturing, with suggestive evidence of input-output linkages. The results show that industrial policy design must account for how market failures interact and how gains are shared along the supply chain