The committee discusses a full resume, including all of the metrics and the actual games.
The more you dig into the actual results on a resume, the more you eventually just wind up at the resume metrics WAB or SOR because they do a pretty good job of summarizing it. So especially at the bubble cutline, where the most thorough discussion takes place, the committee results definitely skew towards the resume metrics. Not necessarily because they use them as the guiding star, but because the process itself is similar to the way the metric is calculated.
But outliers in the predictive metrics and especially big wins or whatever can skew the results in a direction, like UNC being far better than all the teams under consideration in the predictive metrics likely getting them a bid last season when the resume impressiveness according to the resume metrics was otherwise similar (despite the utter terrible performance in Q1).