Experiment arms finer than location error

Assigning experiment arms at a cell grain finer than the delivery channel's location error radius misplaces households across arms. With shared error the misassignment is non-differential (attenuates lift toward null, a power cost); when arms differ in IP-resolvability — multi-family share, carrier NAT, fixed-wireless penetration — the error becomes differential and biases the estimate, not just weakens it.

Detection

Arm-unit equal-area radius smaller than the channel's location error radius; arm-level covariate imbalance on multi-family share or fixed-wireless penetration.

Mitigation

Randomize at a unit whose radius exceeds the worst error radius in the channel mix (R5 for IP-derived delivery); activate finer inside treated units
Block on IP-resolvability and randomize within strata
Never run an IP-derived outcome against IP-derived arms — correlated misassignment can manufacture lift rather than attenuate it

Detection

Mitigation

Affected conversions