Tracking the origin of the accelerating expansion of the Universe remains one of the most challenging research activities today. The final answer will depend on the precision and on the consistency of future data. The sensitivity of future surveys and the control of the errors are crucial. We focus on futur supernovae surveys in the light of the figure of merit defined by the Dark Energy Task Force. We compare different optimisation and emphasize the importance of the understanding of the systematic error level in this approach and their impact on the conclusions. We discuss different representations of the results to distinguish $\Lambda$CDM from other theoretical models. We conclude that all representations should be controlled through combined analyses and consistency checks to avoid some bias.