Here are some rules of thumb:
1) To get good retention of your analytes, one usually needs about 70-75% organic solvent in the case of proteins, 85% ACN in the case of tryptic peptides, and 85-95% for small molecules that are only modestly polar. Exceptions include small molecules that are extremely polar, such as aminoglycoside antibiotics (e.g., kanamycin). To use higher levels of organic solvent than the concentration necessary for good binding risks precipitation of your analytes. For example, complex carbohydrates start to precipitate at 80% ACN, while many tryptic peptides aren't soluble at 90% ACN (but they are at 85%).
2) If you need 90+% to get retention of an analyte, then you usually need to run only to 50-60% organic solvent in a gradient to insure its elution. For a particularly heterogeneous mixture such as a tryptic digest or a crude extract of metabolites from a leaf, some of the components may not elute until you get down to ~ 40% ACN. In such cases it's prudent to run the gradient to 10-20% ACN.
3) With analytes that contain numerous polar binding sites, such as intact proteins or aminoglycoside antibiotics, run the gradient down to 5-10% organic solvent. It may also be necessary to run a simultaneously increasing salt gradient or to add a chaotrope such as 50 mM unbuffered formic acid or hexafluoro-2-propanol (chaotropes antagonize binding in HILIC, the opposite of their effect in RPC).
Hope this helps.