You run a geo experiment yourself. A conversion lift test is run by the ad platform, and it splits at user level, not region level. That means within the same region some people see the ad and some do not, and the platform compares the two groups' conversions.
It is more precise than a geo test, because the split is random. But there are two conditions: the platform requires a minimum spend, and the platform calculates the result itself, which means the same party sits the exam and marks it.