Well, when we get obvious deviations between IQ and g (the thing IQ is stated in the academic literature as what it's trying to measure) some will say that the IQ test was administered incorrectly. Some will say that IQ tests aren't perfect.

IQ tests are probably both not perfect and administered incorrectly (often enough).

I think it's an almost meaningless number unless we're talking about very large groups of individuals, and there's a lot of evidence and theory to support that.