Our ancestry algorithm works by finding long identical segments of DNA between your dog and purebred dogs in our reference dataset. These long identical stretches of DNA are more likely to reflect recent ancestry.
Many dogs descend from other dogs that were themselves mixed breed. The further in the past your dog has purebred ancestors, the smaller the identical segments matching our reference dataset are. In such cases, the best matches to our reference dataset are so small that they can no longer be confidently assigned to any one particular breed. Instead, these are assigned to the "Supermutt" percentage.