well there are two main measures:
1. if writing the same task in another programming language results in much much faster program, then there's much less of a reason for people to use it, as they could do the same task with another language and better performance.
2. a second measure is to compare to existing datalog implementations. the datalog+bdd combination already proved itself to be better, cf. e.g. here https://www.youtube.com/watch?v=-KZY15tItgQ (note that tml is not datalog but the overlapping part should be competitive)