How do you bring field learning back into risk analysis?

When a complaint, service issue or near miss reveals a new risk, how does your team carry that learning into future analysis? What makes the feedback loop effective?