Set Document Name Matching Policy
Start with Balanced mode. Most hotels complete setup in under 3 minutes.
This guide helps you control how strictly AVA compares reservation names with passport or ID names.
Where to Find It
Settings → Security → General Settings → Document Name MatchingQuick Reference
| Setting | What It Controls | Typical Outcome |
|---|---|---|
| Strict | Highest match strictness | More guests go to front desk review |
| Balanced | Default match strictness | Good security with smoother self check-in |
| Lenient | Lower match strictness | More abbreviated names pass automatically |
| Custom | Manual thresholds and rules | Outcome depends on your configured values |
Strict mode accepts omitted patronymic markers like bin, binti, binte, bte, S/O, and D/O.
It still blocks unsafe partial matches.
When both names are available in the same script, AVA compares that native-script text first. That improves matching for Thai and CJK reservations without changing your selected mode.
When names use different scripts, AVA also checks deterministic transliterations for token coverage. This prevents one short transliterated token from approving an incomplete lenient match. Exact native-script matches remain accepted.
Two-Token Document Names
When a document has two tokens, AVA requires meaningful support for both tokens.
This guard applies to Balanced, Lenient, and Custom modes. It prevents one token from carrying the whole match by itself.
AVA still accepts compact or swapped-token matches when both document tokens are represented.
Surname-Only Matches
When both name parts are available, a shared surname cannot approve the match by itself.
This protects guests who share a compound surname, such as DELA PAZ.
Valid abbreviations, aliases, name reversals, and native-script evidence can still pass when both document tokens are supported. The existing ACI staff override remains available when staff verify the guest in person.
| Name Pair | Result | Why |
|---|---|---|
| "Alex DELA PAZ" vs "Jordan DELA PAZ" | ❌ Review | Only the shared surname supports the match |
| "Zilin Zhu" vs "Zhuzilin Zhuzilin" | ❌ Review | One document token is not meaningfully supported |
| "A B" vs "B A" | ✓ Pass | Both tokens are present, even if order changes |
| "M Taka" vs "Mitsui Takafumi" | ✓ Pass | Both document tokens still appear in the match |
Set the Matching Mode
-
In Document Name Matching, choose a mode
-
Review the scenario preview cards
-
Click Save General Settings
✓ Your selected policy applies to new check-in attempts.
The preview cards now use the standard curated scenarios. Strict, Balanced, and Lenient show the same example list. The outcomes follow CCI's token-level scoring and two-token safety guard.
Tune Custom Mode
Use Custom mode when your team has a defined review SOP.
- Choose Custom
- Set Base similarity threshold (80-95)
- Choose whether to allow:
- Initial + surname prefix matching
- Alias name matching
- Set abbreviation controls:
- Minimum surname prefix length (3-6)
- Minimum surname coverage (0.30-0.80)
- Require two tokens
- Require single-character initial
- Click Save General Settings
Automatic Safety Rules
AVA normalizes document matching values when you save:
- Invalid mode values reset to Balanced
- Custom threshold must stay between 80-95
- Minimum surname prefix length must stay between 3-6
- Minimum surname coverage must stay between 0.30-0.80
"true"and"false"text values are converted to real toggles
✓ If you enter out-of-range values, AVA saves corrected values.
Patronymic Names
| Name Pair | Strict | Balanced | Lenient |
|---|---|---|---|
| "AHMAD BIN IBRAHIM" vs "AHMAD IBRAHIM" | ✓ Pass | ✓ Pass | ✓ Pass |
| "AHMAD IBRAHIM" vs "AHMAD BIN IBRAHIM" | ✓ Pass | ✓ Pass | ✓ Pass |
| "AHMAD BIN IBRAHIM" vs "AHMAD BIN ABDULLAH" | ❌ Review | ❌ Review | ❌ Review |
Malay Common-Name Omissions
Use this for one approved leading Malay token only. It works in Balanced and Lenient modes.
Sometimes a Malay name includes one leading token that AVA can safely omit. AVA still requires the remaining tokens to form a strong match.
This keeps a shared Malay prefix from passing on its own.
Strict and Custom modes do not use this exception. AVA applies the same matcher in reservation lookup, document upload, staging suggestions, and staged commit validation.
| Name Pair | Strict | Balanced | Lenient |
|---|---|---|---|
| "MOHD AHMAD BIN IBRAHIM" vs "AHMAD BIN IBRAHIM" | ❌ Review | ✓ Pass | ✓ Pass |
| "MOHD AHMAD BIN IBRAHIM" vs "MOHD AHMAD BIN IBRAHIM" | ✓ Pass | ✓ Pass | ✓ Pass |
| "MOHD AHMAD BIN IBRAHIM" vs "MOHD AHMAD BIN ABDULLAH" | ❌ Review | ❌ Review | ❌ Review |
| "MOHD AHMAD BIN IBRAHIM" vs "AHMAD ABDULLAH" | ❌ Review | ❌ Review | ❌ Review |
Common Examples
| Name Pair | Strict | Balanced | Lenient |
|---|---|---|---|
| "John Michael Doe" vs "John Michael Doe" | ✓ Pass | ✓ Pass | ✓ Pass |
| "Jon Smyth" vs "John Smith" | ❌ Review | ✓ Pass | ✓ Pass |
| "M Taka" vs "Mitsui Takafumi" | ❌ Review | ✓ Pass | ✓ Pass |
| "K Taka" vs "Mitsui Takafumi" | ❌ Review | ❌ Review | ❌ Review |
| "A Kumar" vs "Arun Kumar" | ❌ Review | ✓ Pass | ✓ Pass |
Alias Names
An alias can pass only when all three conditions are true:
- Alias name matching is enabled.
- The ID supplies the alias.
- The alias matches the booking under the selected mode.
Different primary surnames alone do not pass.
| Booking | Primary name on ID | Alias supplied by ID | Balanced result |
|---|---|---|---|
| "Alice Tan" | "Alice Lim" | "Alice Tan" | ✓ Pass when alias matching is enabled |
| "Alice Tan" | "Alice Lim" | None | ❌ Review |
| "Alice Tan" | "Alice Wong" | "Alice Chen" | ❌ Review |
| "Alice Tan" | "Alice Lim" | "Alice Tan" | ❌ Review when alias matching is disabled |
Lenient mode can reduce front desk workload, but it also accepts more borderline name patterns.
Troubleshooting
Too many guests are sent to review
What you see: Many valid guests are redirected to staff review.
Fix:
- Switch Strict to Balanced
- In Custom, lower threshold slightly
- Enable abbreviation matching if OTA names are often shortened
Too many abbreviated names are passing
What you see: You want tighter screening for short names.
Fix:
- Switch Lenient to Balanced or Strict
- Increase threshold in Custom
- Increase surname prefix length or coverage
Changes do not appear to apply
What you see: Results look unchanged after editing.
Fix:
- Click Save General Settings again
- Refresh the page
- Test with a new check-in flow
Custom values changed after saving
What you see: A saved custom value looks different from what you entered.
Fix:
- Re-open Document Name Matching
- Compare your values against allowed ranges
- Use values inside the range and save again
Still Stuck?
Contact success@vouch-technologies.com if:
- ❌ Name matching results stay inconsistent after retries
- ❌ Settings save successfully but behavior does not change
- ❌ You need help tuning custom values for local compliance rules
Helpful details to include:
- Your selected mode
- Custom values (if used)
- 2-3 example reservation vs document name pairs