A marketing analyst is using a large language model to sort customer reviews for a new smartphone. They need to classify each review as 'Positive', 'Negative', or 'Neutral' without providing the model with any pre-labeled examples. The analyst's current prompt is failing to produce consistent results. Analyze the provided prompt and case details, explain why the prompt is ineffective, and then construct a revised prompt that would be more successful for this task.