One thing that doesn't show up in a screenshot: the anomaly detection isn't just "biggest number wins."
It looks for things like sudden spikes, broken patterns, or values that don't fit the trend in your dataset the kind of thing you'd normally catch by eyeballing rows in a spreadsheet, except most people don't have time to actually do that every week.
Happy to answer questions about how it works under the hood, or what triggered building it this way instead of a simpler threshold-based flag.