You have a binary classifier that outputs probabilities, and the team needs a decision threshold for turning scores into actions. Different thresholds change precision and recall, so the right choice depends on the business cost of false positives and false negatives.
How do you choose an operating threshold for a classifier?