Creating a floxed gene is simply adding loxp recombination sites to a particular portion of the the genomic DNA you wish to eliminate (through cre-mediated recombination). There are likely factors epigenetic factors (such as folding) that may influence accessibility to the recombination sites, but overall it should not matter how the gene is regulated at the transcriptional level to get recombination and a good KO of your gene.
if the floxed gene is huge, such as >20kb, the Cre recombination efficancy sometimes won't be great and you will still have some expression. However, if your target gene is small, the floxed sites and Cre activity won't be an issue later. If you have to KO a big gene, maybe you can flox it at promoter plus its 1st exon region.
I think the first question one should ask before knocking out a gene is whether or not the gene is essential or not for cell viability.
One should really be careful when knocking out only part of a gene. DNMT1 knockout in cells is a good example of that. For years people have consider it as a null mutant but later on it turned out to be a hypomorph version. The strategy involved deleting exons 2-5 while putting out of frame the rest but somehow cells have managed to make a functional version which allowed them to survive and they shouldn't..
Thanks Abdelhalim, that is a good point. Ive actually floxed CD40, which was difficult to make. The knockout mouse exists as a partial knockout. That works pretty well functionally. The protein is expressed but altered in function. I thought I remembered someone saying that floxing only works if the gene is activation induced. As I thought more about it, that was the efficacy of siRNA. Thanks again.