Thanks!
It actually does 3 different comparisons... namely, we do a pixel-by-pixel comparison, a DOM comparison, and a "content" comparison (i.e., the words that the end-user sees).
We only trigger a new version that we alert the customer about, however, when actual content changes. We provide the pixel-by-pixel screenshot, too (but it's often not very useful unless the changes are very small and image based). (We also don't currently show the DOM comparison as we assume it won't be valuable for most of our target customers.)
