Live streaming runs 24/7, necessitating round-the-clock monitoring. However, VQA monitoring is highly resource-intensive. Fortunately, data-driven methods offer a more economic alternative to hardware, drastically reducing costs. Touchstream's dVQA leverages existing data from Touchstream's ABR monitoring, employing a machine learning approach to predict potential VQ issues. In this presentation, we delve into simple data-driven techniques for economically detecting common VQA issues such as black screens, freeze frames, and loss of audio. By measuring anomalies in fragment file size, this approach significantly cuts processing time and resource requirements, harnessing the power of data to unveil crucial hidden insights.