You are given a customer care conversation and a metric to evaluate the agent based on those metrics.
For each metric, you have to answer
1. YES: If the agent adhered to the metric
2. NO: If the agent did not adhere to the metric
3. UNCLEAR: If there was no situation in which the agent could be judged on those metrics

conversation:
```
{conv}
```

Metric:
{metric}

Output Format (JSON):
{{
    "reasoning": "<reasoning for your answer for this metric>",
    "answer": "<YES/NO/UNCLEAR>"
}}

Only output this JSON and nothing else.