Nope, not the way I used it in the trial at my uni. It’s a “similarity score” indicating potential plagiarism. Turnitin is only a helper. It does not and cannot detect plagiarism any more than a fingerprint matcher can detect a murderer. The problem is that the definition of plagiarism we’re using is too loose. Universities in their rules have very clear definitions.
Similarity is like fingerprints — their presence at a crime scene doesn’t necessarily implicate the persons who left them as a perpetrator. You have to examine the entire case. There may be good reasons for similarity (e.g., proper citations).
Plagiarism (as well as its sanctions) have definitions that vary by university. My university doesn’t call it plagiarism until work has been submitted. We want to use Turnitin on draft dissertations before students submit them as final, so that they correct any issues with similarity (or are ready to explain them when the committee gets the results). The whole idea is to AVOID plagiarism (as a formal academic act recognized by the university which has sanctions). When it gets to the sanction level, it’s costly to everyone (the $ amount is way more than Turnitin is making, in my opinion).