Understanding Match Confidence Levels

Every match result includes a confidence level to help you make informed decisions. Here's what each level means and how to handle them.

Confidence Levels

HIGH (Score >= 0.80)

Very confident match. Safe to accept automatically in most cases. The hotel names are nearly identical or clearly the same property.

Examples:

• "Hilton Paris Opera" → "Hilton Paris Opera Hotel" (0.95)
• "Marriott Times Square" → "New York Marriott Times Square" (0.88)

MEDIUM (Score >= 0.60)

Likely a correct match, but review is recommended. Common with abbreviated names or slight variations.

Examples:

• "Grand Hyatt NYC" → "Grand Hyatt New York" (0.72)
• "The Ritz" → "The Ritz-Carlton London" (0.65)

LOW (Score >= 0.54)

Possible match, manual review suggested. The names share similarities but may be different properties.

Examples:

• "Hotel Paris" → "Paris Hotel & Casino Las Vegas" (0.56)
• "Grand Hotel" → "Grand Hotel Europa" (0.55)

NO_MATCH (Score < 0.54)

No confident match found. The hotel may not exist in our database or uses a significantly different name.

Understanding Unmatch Reasons

When a hotel doesn't match, we provide a reason:

Reason	Description
NO_CANDIDATES	No similar hotels found in our database
BELOW_THRESHOLD	Candidates found but none scored above 0.54
AMBIGUOUS_MATCH	Multiple hotels scored similarly (rare)
INVALID_INPUT	Missing hotel name or malformed data

Best Practices

1. Include location data - City and country improve accuracy significantly
2. Use full hotel names - Avoid abbreviations when possible
3. Review MEDIUM and LOW matches - Set up a review queue for these
4. Track NO_MATCH hotels - They may need to be added manually

Automating Your Workflow

Based on your business requirements:

High Automation (Accept HIGH only)

• Accept: HIGH matches automatically
• Review: MEDIUM and LOW manually
• Reject: NO_MATCH

Moderate Automation (Accept HIGH + MEDIUM)

• Accept: HIGH and MEDIUM automatically
• Review: LOW manually
• Reject: NO_MATCH

Manual Review (All)

• Review: All matches manually
• Best for critical applications

Choose the approach that balances accuracy with operational efficiency for your use case.