An AEO experiment, and what a small test can and cannot tell you

Answer engine optimisation advice is easy to give and hard to verify. The useful move is to run a small controlled test: take a set of pages, apply the AEO changes to half, leave the other half alone, and watch citation behaviour over a few weeks.

A write up of exactly that kind of experiment is here, with a Ukraine market version covering the same method.

Two cautions when you read any test like this. Sample size is usually small, so treat the direction as a signal and not a law. And answer engines change their behaviour between runs, so a result from August may not hold in October. The value is in the method, not the exact numbers: define the change, hold everything else constant, and measure citations rather than vibes.