You should be using a list of tuples here:
>>> lst = [d.items()[0] for d in similarity]
>>> lst
[('Ben', 49), ('Moose', 18), ('Reuven', 39), ('Cust1', 58), ('Cust2', 10), ('Francois', 58), ('Jim C', 39), ('Iren', 13), ('Cust3', 13)]
Then you can sort those as usual.
>>> from operator import itemgetter
>>> sorted(lst, key=itemgetter(1))
[('Cust2', 10), ('Iren', 13), ('Cust3', 13), ('Moose', 18), ('Reuven', 39), ('Jim C', 39), ('Ben', 49), ('Cust1', 58), ('Francois', 58)]
If you want, you can also use a single, ordered dictionary to hold the values:
>>> from collections import OrderedDict
>>> OrderedDict(sorted(lst, key=itemgetter(1)))
OrderedDict([('Cust2', 10), ('Iren', 13), ('Cust3', 13), ('Moose', 18), ('Reuven', 39), ('Jim C', 39), ('Ben', 49), ('Cust1', 58), ('Francois', 58)])