Anything which is correlated to IQ, certainly including academic achievement, provides some information about IQ. So if someone is doing better/worse in school than his score on an IQ test predicts, you should revise your estimate of his IQ up/down, although I can't say by how much.

Here is a thought experiment. For a group of students, you have IQ scores from the Stanford-Binet (SB) and WISC, taken at the end of 8th grade, in addition to middle school (grades 6-8) grade point average. I bet that linear prediction of WISC IQ from SB IQ and GPA would have positive coefficients on both variables.