Which kind of graphid have you used, and how did you make the composites ? Are you sure that the homogeneity was good ? It depends on the kind of graphite, but 15 wt.% may be already quite high.
The thinner graphite is the better its conductivity. So your best bet would be to seek to use the thinnest graphite (single layer with ring irregularities) that affords best uniform mixing (homogeneity). All this must be ascertained with empirical evidence and not based on whim or intuition. Experiments should be done to determine the limits of proportions of graphite that affords the best thermal stability.