Good data starts long before anyone opens a tablet. It starts with a field team that understands the community, a questionnaire that has been piloted and revised, and a plan for what happens when the network goes down in a village with no signal.
Over ten years of data collection across Tanzania, a few lessons keep proving themselves true. First, local knowledge is not optional context — it is the difference between a response rate of 60% and one of 95%. Second, every data quality check you build in at the point of collection saves you ten hours of cleaning later. Third, a happy, well-briefed enumerator produces measurably better data than a rushed one.
We also learned to design for the field conditions we actually have, not the ones we wish we had: offline-first data collection tools, solar chargers in the car, and a supervisor who reviews a sample of submissions every single evening. None of this is glamorous. All of it is what separates research that holds up from research that does not.