A consultant configures conversation intelligence for a financial-services sales team, adding custom tracked keywords tied to a compliance disclosure that must be spoken during any call discussing investment products. During a live call, the AI flags that the required disclosure phrase was never spoken. What should the sales team do with this signal?

The correct answer is C. Keyword tracking can miss required language when a rep phrases the disclosure slightly differently than the tracked term, so the flag is best treated as a prompt to review the actual call and confirm independently whether the disclosure was genuinely omitted, rather than as a final determination either way. Option A is incorrect because it treats automated keyword detection as infallible and skips the verification step, which risks escalating a false positive as if it were a confirmed violation when the rep may have used acceptable alternate wording. Option B is incorrect because it sets an arbitrary threshold, waiting for three occurrences, that has no basis in the scenario and would let a genuine single compliance gap go unaddressed while waiting for repetition that may never come. Option D is incorrect because it overreacts to one instance by abandoning the tracking capability entirely, discarding a useful early-warning signal rather than simply verifying and, if needed, refining how the phrase is matched. The appropriate response keeps the human review step in the loop rather than trusting or dismissing the AI signal outright.