It depends on the purpose for the measure.
For example, if we measure your tinnitus loudness and via protocol we conclude, with your "subjective assessment", that it is equivalent in loudness to a signal that is 30 dB loud, then this is what you perceive it to be.
For all practical purposes, this is what you feel it sounds like, and that's all that matters to you, whether that's due to a 1 nanoWatt signal on an auditory pathway or a 10 nanoWatt signal - it doesn't matter for this purpose (it does matter when we want to figure out a mapping between electrical power and tinnitus volume, but this is a rare endeavor: in practice, we want to help patients perceive a less loud tinnitus).
If we measure it again post treatment and you claim that it now is 20 dB, then from your perspective, which is all that matters to you, you've actually perceived a reduction of 10 dB.
So for this particular purpose, the subjective nature of the measurement is irrelevant.
However, if we wanted to tie a disability payment to a volume level, we'd have to think twice about using a patient reported measurement, as you can imagine it can easily be gamed. For those purposes, measurement techniques that don't require patient feedback make sense.