Typically, I get the numbers like Dan does, from the NMR integration. Especially if the chromatography is separating two compounds, I know the identity of both compounds, and I have mixed fractions. No, the numbers don't make for good publications, but when you're just trying to figure out how well the reaction worked and if there is enough product in your mixed fractions to make it worth chasing, it's worth the effort.
If you truly want good accurate numbers, you can do it by running standard curves on an HPLC or GC for your pure compound, then testing your impure compound. The same way you would be quantifying a compound in a solution of unknown concentration. Make sure the concentrations of your standard samples are on both sides of the concentration of your mixture so you can interpolate instead of extrapolate from the curve.