> ## Documentation Index
> Fetch the complete documentation index at: https://docs.dataspike.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Name Matching & Fuzziness

> How AML Search handles typos, transliterations and spelling variants, and how to choose the right fuzziness level.

Names rarely appear the same way twice. They are mistyped, transliterated from other scripts, shortened or written with and without a middle name. **Fuzziness** controls how far a listed name may differ from your query and still count as a match — so a spelling difference doesn’t hide a real hit.

## Fuzziness levels

Fuzziness is set per search, from **Exact Match** to **Broad Match**:

| Level | What it matches |
| - | - |
| **Exact Match** | Only names spelled exactly as entered. No tolerance for typos. |
| **10% – 90%** | Increasing tolerance: the higher the level, the more spelling variations are accepted. |
| **Broad Match** | The widest setting: 1 to 3 misspellings per word, depending on the word’s length. |

> [!NOTE]
>
> **Broad Match is the default.** It catches the most spelling variants, which matters most for transliterated names — and it also returns the most false positives. Lower the level if your analysts spend too much time clearing irrelevant hits.

### What Exact Match still finds

Exact Match switches off tolerance for **spelling** differences, not for the different names a record carries. A listed record often stores several versions of a person’s name — in other scripts, in other transliterations, with a middle name or patronymic. Your query is compared with each of them, so an exact query can still return a record whose main name looks different.

## Example: searching for “Nikolaos Papadopoulos”

The same query at three levels:

**Broad Match**

Returns the widest range of possible matches, for example:

* Nikolaos Papadopoulos
* Nikolaos Papadopulos
* Nikolas Papadopoulos
* N. Papadopoulos
* Papadopulos *(surname only)*

Best when the input may contain typos or a non-standard transliteration.

**50% fuzziness**

Allows moderate variation:

* Nikolaos Papadopulos
* Nikolaos Papadopoulus

Stricter than Broad Match but still tolerant of common typos — a balance between catching variants and limiting false positives.

**Exact Match**

Returns only names spelled exactly as entered, across all versions stored on a record:

* Nikolaos Papadopoulos
* Νικόλαος Παπαδόπουλος *(original Greek-script name stored on the same record)*
* Nikolaos Georgiou Papadopoulos *(full name including the father’s name)*

The strictest setting. It will miss a listing whose spelling differs from your query.

## How close is the match

Each result also gets a **match score** showing how closely the listed name matches your query. Missing name parts and spelling differences lower it; exact matches on individual words raise it. For a listed record named **Nikolaos Papadopoulos**:

| Query | Differences | Match score | Why |
| - | - | :-: | - |
| Papadopulos | First name missing; surname spelled differently | below 50% | Both the missing first name and the surname differences lower the score. |
| Papadopoulos | First name missing | \~50% | The surname matches exactly, but the missing first name holds the score down. |
| Nikolaos Papadopulos | Surname spelled differently | 75 – 80% | The exact first name lifts the score despite the surname differences. |

The match score is separate from the [risk score](/aml-screening/how-risk-scoring-works): it says how likely the record is the person you searched for, not how risky that record is. Only strong name matches raise the risk level of the overall search.

## Choosing a level

| Situation | Recommended level |
| - | - |
| Onboarding and periodic screening, where a missed hit is costly | **Broad Match** (default) |
| Too many irrelevant hits on common names | **50% or lower** |
| You have the exact legal spelling, e.g. from a verified passport | **Exact Match** |

> [!TIP]
>
> To cut noise on common names, try adding a date of birth or country to the search before lowering fuzziness: filters narrow the results without making a misspelled listing harder to find.

## Next steps

> **[AML Search](/aml-screening/search)**
>
> Run a search in the Dashboard or via the API.

> **[How Risk Scoring Works](/aml-screening/how-risk-scoring-works)**
>
> What the score on each match means.

> **[Daily Monitoring](/aml-screening/monitoring)**
>
> Re-screen applicants automatically every day.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.