I wonder if a dilution series and line equation of (Me-C 23:0) Methyl Tricosanoic Acid as internal standard(I.S) and using it for quantitation of
C 4:0 till C 20:0 in milk is fairly free from errors in terms of accurate recovery calculation.
From analytical point of view C23:0 could not be so good for splitless injection due to less volatility respect to low boiling methyl esters and discrimination can be present. Anyway you have not many more choice with milk fat. C13:0 is present at 0.15%, C15:0 at 1.2 %, C17:0 at 0.66%, C19 and C21 are easily overlapped with C18 unsaturated and anyway present at least to 0,5%. C23:0 is present into the milk fat between 0,01 to 0,02 percent.
As alternative you can add after methylation another volatile substance that elute into middle part of chromatogram such as a ethyl ester of C12 or 14. This is not a true ISTD but can correct the GC injection.