TFI is so problematic that I largely discount it as a metric. But let me put a bit of a narrative on these numbers and theorize why that 15-20 point reduction in TFI is all we see from a 75% reduction in tinnitus loudness.
First of all, that reduction in TFI represents roughly a drop in one class of tinnitus distress; "not a problem (0 to 17), small problem (18 to 31), moderate problem (32 to 53), big problem (54 to 72), very big problem (73 to 100)"
If the average participant came in around a 50, they dropped from a borderline big problem to a borderline small problem.
This isn't insignificant. But again TFI is a really poor way to measure tinnitus, it measures our emotions about tinnitus. Though there is a life changing functional difference in a 75% reduction in tinnitus loudness for severe tinnitus sufferers (hearing, concentration, relaxation, lack of a pain response), an average person with average tinnitus in the study might still express irritation that they can hear it in a quiet room or that when they go to bed at night they need to put on a fan and the TFI captures their annoyance they need to listen to this sound at all. In short TFI, the measurement of how we feel about tinnitus, probably isn't linear, we can greatly reduce the volume yet still be troubled that there are phantom sounds at all.
Secondly, maybe the 30 weeks wasn't enough time to capture the habituation process for the new quieter tinnitus. I think, as a rule of thumb, we normally think of someone with 'normal tinnitus' taking 9-12 months to really get over the hump.