ELISA is a semi-quantitative technique, the absorbance reading depends on a lot of factors like time of development, concentration of secondary antibody and activity of the linked enzyme, temperature, efficiency of protein binding to the plate, ...
You can make it quantitative by having some wells with known antigen concentration on the same plate, to generate a standard curve. I'd recommend using that curve only for samples incubated on the same plate, at least until you have standardised your protocol so that the standard curve from several plates reliably superimpose.
With crude preparations it is not possible to give the concentration of standards in absolute units like µg/mL or µM, in that case one uses the biological effect to measure concentrations. For example, for enzymes the international unit was defined as µmol substrate turned over per min. Nowadays one should use the SI unit katal instead (mol/s). For vitamins, before their structure was known, units based on physiological effect (ability to cure a hypovitaminosis) were also used.