Designing scoreboards is probably the most complex activity when building a testbench. The scoreboard keeps track of various transactions across the entire system and sign-off the correctness of the behavior.
As an example, take a NOC. If a transaction is driven on interface A at clock cycle N, and it is expected that the NOC will generate a new transaction at clock cycle N+2, on interface B, then a scoreboard that gets all the data from interface A and interface B can ensure that the DUT is behaving as expected. So in general, answer to your question is No.
In some limited corner cases, may be okay to restrict the scoreboard to monitor transactions of a single interface. This is true if all the input and response transactions happen on the same interface.
Logie Ramachandran
Accelver Systems Inc.