Unless I'm misunderstanding the question, you need a standard curve for the products of the reaction, not the substrate. Then you can convert absorbance for your samples into the amount of product generated, by extrapolation from the standard curve.
It would still be better to measure product, unless there is a crazy number of them. With substrate, you will have non linear behaviour if you consume more than 10-15% of substrate. This means you are forever subtracting a large number from a large number to get the amount of substrate consumed, and poor assay precision may be anticipated.