Sentiment at Scale

Classify a batch of customer reviews by sentiment — using a 5-point scale and explaining the reasoning for edge cases.

Skills Tested

Sentiment analysisConfidence calibrationStructured output

Challenge Prompt

This is the example prompt structure. Write your own variation below.

Classify each review below on a 5-point sentiment scale: 1=Very Negative, 2=Negative, 3=Neutral, 4=Positive, 5=Very Positive. Output as a CSV table: Review_ID, Score, Confidence (High/Medium/Low), Reason (one word). For any score of 3 (neutral), add a two-sentence explanation of the ambiguity. Reviews:
1. "Delivery was late but the product itself is excellent."
2. "Fine I guess."
3. "Absolute disaster from start to finish."
4. "Exactly what I expected — nothing more."
5. "Life-changing. I tell everyone about this."

Your Prompt

Write your own version. Every word matters — be specific.

🪙

Reward

+100 GeraCoins

Earn GeraCoins for completing this challenge. Redeem across all Gera products.

Hints (3)

1Confidence scores surface the AI's uncertainty — use them to flag reviews for human review.
2The neutral-explanation rule forces the AI to articulate mixed-signal reasoning.
3CSV output makes bulk processing straightforward.