There will be random error in each of the intensity measurements due to the strength of the signal. The standard deviation in each channel's counts will be the square root of the count. The count qualifies as a Poisson distribution. You usually get a good sense of that from the raw pattern. For short acquisition times, there is a lot more noise in the signal than for longer times. Ultimately, that will be a random part of the error that you are looking for.
However, for XRD, there is a lot of question about the goodness of the fits and the quantitation. First, you have to have the phases properly identified beyond question. We had one application with Portland cement mortar which failed to identify the mineral dolomite. Leaving it out made for a serious error in the other values. Another case identified a phase as MoC when it was really TiC. They have very similar patterns. Both were plausible, but another technique was needed to identify TiC as the true phase.
I might attempt quantitative XRD if everything is known and the only difference is the treatment. However, that is quite rare. Instead, I think there is great value in qualitative comparison. If there is a significant difference between samples, it should show up there.
In cases like this, it is always good to have the specific figure or table to refer to. Please share a copy.
Maybe I would have the same question. Or maybe I would find the reviewer's question to be spurious. It seems like some pose critiques just for the sake of challenging something.