That depends on a lot of factors, and the best approach would probably be to perform a serial dilution of your sample on your first run. You should perform this dilution for each new assay you are preforming, as different targets can be at very different abundances (ex. GAPDH vs lncRNAs). This is a simple approach to determine what dilution to use in your experiment. This will also help you to clearly establish the amplitudes of your positive and negative droplets (in addition to your no-template control), as it can be confusing to delineate positive and negative droplets when high template concentrations are present. It will also help you assess the sensitivity of your assay. Until you perform an initial experiment, it will be difficult to know what dilution to use. Often times the first run with a new assay may be worthless if you go into it blindly adding an arbitrary amount of template, so performing a dilution to start with can save you money and time (time especially if you're making droplets manually!).

Good luck!