A new tannin from the fruit extract of T. chebula has been located in 1-butanol fraction in two (or probably more) isomeric forms. The interesting peaks are weakly retained and poorly separated in PR-18 column with acetonitrlie/1%HCOOH gradient 0-25% in 15 min at 1ml/min. It shows the hydrophoibic nature of the compound.
To get base line separation and K >0.5min, I have tried different dradient profiles and with CH3CN, mixtures of CH3CN and MeOH, and MeOH using 0.02-5% HCOOH and 0.05M H3PO4 with no luck:
Now
Q1. Sould I try some other solvent profiles?
Q2. Should I try NP column instead of RP?
Q3. Should Itry LIQ-LIQ extraction what should be my choice?

If they are that weakly retained, I would try to run the extract isocratically in water with 1% HCOOH. If they are not retained enough then, there are two things that you can try.
1. A Symmetryshield RP18 column: this column gives up to 3x higher retention than a C18 for phenolic compounds, and might do the trick for the tannins as well.
2. If this is not good enough, you can try hydrophilic interaction chromatography. Behind this is nothing but normal phase chromatography with a very polar organic solvent ant water in the mobile phase. Use a silica column with acetonitrile and 10% water, if nothing comes out, increase the water content. As a reference: in something like this, sugars elute between 20 and 30% water.

I can add another solution of what Uwe proposed. You can use a Porous graphitic carbon column (eg Hypercarb from Hypersil). Due to its structure, you have strong pi-pi interactions (sorry but the forum doesn't support Greek caracters) and normaly tannins will be very retained. Furthermore, this column allows in general higher selectivities of isomers (not enantiomers) than conventional RP18 columns.