Haptic shared control is a driving assistance system that allows for continuous communication between the driver and the automation through a physical control device, such as the steering wheel. Previous research has proposed the Four-Design-Choice-Architecture, a control system with a feedforward loop that opens possibilities for personalisation to the driver. The goal of this research is therefore to understand how changing the reference trajectory and the feedforward gain influences the acceptance of the system. This investigation consists of an offline simulation platform and a human-in-the-loop experiment in which the most popular driving group, R3L5 drivers, and the optimal curve cutters group, R2L1 and R2L2 drivers, were subjected to four different reference trajectories and two feedforward gains. The reference trajectories consisted of the industry standard centerline trajectory, a completely personalised trajectory and the class-average trajectories of the drivers' own class and the other group's class. The feedforward gains consisted of the heuristically tuned value of 0.92 and the lower 0.5 to give the drivers a higher workload. The results showed that complete personalisation consistently leads to less conflict for all drivers, however, drivers that adapt their driving style to the guidance also reach low conflict values for the class-average guidance of their own class. Furthermore, it was found that acceptance is not linked to optimal trajectories, i.e., less steering input needed and more dynamically safe trajectories. The results also showed that the higher feedforward gain of 0.92 is beneficial with respect to the lower gain as it reduces conflicts up to 86.91\% and is rated consistently higher in subjective questionnaires. Future research should focus on understanding which drivers need the complete personalisation and on how to optimise the feedforward gain.