How do geo-based experiments work?

Geo-based experiments are designed to measure the causal impact of a marketing intervention by minimizing bias. The core principle is to randomly assign distinct geographic regions to either a control group (no intervention) or a treatment group (intervention applied). This random assignment ensures that, on average, the groups are comparable before the experiment begins, allowing any observed differences in outcomes to be attributed to the intervention. The regions must be non-overlapping to prevent "spillover" effects, where the treatment in one region might influence sales in an adjacent control region, thereby contaminating the results. Overlapping regions would make it impossible to isolate the true effect of the treatment. Assigning regions only to a control or only to a test group would not allow for a comparative analysis.