Anyone done farmer segmentation research before — where do you even start?

Yeah, demographics alone will mislead you badly with smallholders. We've had way better luck segmenting on decision-making behavior and risk orientation — things like whether they trial new inputs before committing, how they weigh yield vs. cost, and who they trust for advice (neighbor vs. agro-dealer vs. extension officer). A quick attitudinal survey with cluster analysis on those dimensions usually surfaces 4-5 meaningful segments that actually map to different marketing approaches.