This is likely just liquid class optimization for specific lab/liquid - but multidispensing is a difficult technique to optimize. The simplest solution here would be to aspirate ~100µL more than require and dispense the first 2-3 aliquots back into the reservoir if your observations are that only the first 2 alqs are affected.
Beyond that, it’s troubleshooting and optimization. Skim over literature for suggestions, such as Tecan’s LC guide, here’s the link from Tecan with multi pipetting in/around section 4-5 and the Hamilton equivalent (still amazing guidance despite different instrumentation) is here - Library Downloads (Links)
Lab Automation Forums/Hamilton Guides/eBook_Liquid Handling Reference Guide_Revision2_Final.pdf