Each entry in the result shows two parameters, count and distance. Count is the number of reference images in the database which were considered to be close to the input image. Distance is the degree of difference between the input image and average of those reference images.
Basically, higher counts mean higher confidence for the result and lower distances mean closer matchings. In the result page, while entries are shown in the order of counts, it is also worth paying attention to distances.
0.35 or less : Very close. Probably the same person.
0.35 - 0.38 : Close. You can claim that you are a look-like.
0.38 or more : Not very close. Some people might find them resembling.
2 or less : Results are not highly reliable.
3 or more : Results are reliable