Full factorial design

Testing every combination of every factor's variants at once, so both main effects and interactions are measurable.

A full factorial design builds every combination of every factor's variants and randomizes traffic across all of them: two factors with two variants each is a 2×2, four cohorts; three factors with three variants each is a 3×3×3, twenty-seven. That's what buys you a main effect for each factor on its own and an interaction effect between them, the thing a sequence of A/B tests structurally cannot produce, because it never separates the factors from each other in the first place.

The cost is traffic: each additional cell both dilutes the traffic available per cell and adds another comparison that has to clear a stricter statistical bar. See A/B vs. multivariate testing, explained for the sample-size math and a decision framework by traffic volume.