This is a complex issue. Maybe this web site can be usefull for you: http://bio.slu.edu/mayden/systematics/bsc420520lect8.html
Regarding cryptic species, in fact t is problematic, maybe impossible, to find morphological differences. So you must work with differences found by molecular data.
One has to be familiar with the range of individual variation within the species. Cryptic species are usually distinguished on the basis of genetic differences although there are several that are distinguished by differences in internal organs (eg. genitalia in insects). There is no rule of thumb for using morphological characters to distinguish species. Rather, when one has ascertained the range of individual, seasonal and geographical variation within the species, then one can look for features that remain constantly different from closely related species and use these to distinguish them.
The best is to have larger series of individuals. Then, you will be able to recognize characters, that varies within series and those variing among series.
There is no propery that distinguishes variation within a species from variation among species.
If you have good molecular indications, that two distinct Clusters are different species, you can search for morphological characters which mirror the Separation you found.
Ideally this should be testable:
If you then find an individual that you determine to one of the given species, based on your proposed characters, a molekular Analysis should result in clustering in the right Group (test of congruence).
Generally we are too fastly speaking of cryptic species (which by the way are not cryptic if you find some morphological seperation - in the strict sense. Because we have to separate instances were we did not find any characters to differentiate among newly found entities, from such instances where there are no characters in a given System - e.g. morphology - to seperate two entities). Often it is better to firstly speak of overlooked diversity (so using more neutral Terms)