If a single mutation has an important effect, as the mutation near lactase does, then it does. What do you find surprising – that a single mutation in the genome could have a large phenotypic effect, or that a single mutation in the population could spread to everyone?
Sure, there’s extensive theoretical modeling of all of these processes; that’s the subject of the field of population genetics.
There are several possibilities. There can be a single mutation that is successfully selected for; that’s what happened in lactase persistence in northern Europeans. If the beneficial variant isn’t already present in the population when a selective pressure arises, this is probably the most likely way that selection will work, since selected alleles typically reach high frequency quickly compared to the rate of new mutations. A second mutation doesn’t usually have time to occur.
On the other hand, there can also be multiple, identical mutations that occur in different individuals in the same population. That’s unusual, but it does occur, for example in the case of the sickle cell trait. It’s probably relevant that balancing selection is at work in that case, maintaining the allele at an intermediate frequency for an extended period.
There can be multiple mutations with the same phenotype. That’s what happened at lactase worldwide: different, independent mutations confer lactase persistence in Europe and in Africa. In this case, the subpopulations were more or less isolated from one another, and so the first beneficial allele didn’t spread to the second region, leaving time for a different mutation to occur. (Obviously, the probability of this happening depends on how many different mutations would yield the same phenotype. That can vary enormously.)
There can be selection for variants that are already present in the population, so-called selection on standing variation. That’s what happened in the case of selection for greater stature in northern Europe (or perhaps for shorter stature in southern Europe); in this case, hundreds of existing variants, all affecting height, were shifted modestly in frequency, resulting in an overall change.
Well, mutation is a stochastic process, at least at the level we’re ever likely to study it, so that’s bound to be the case.