In addition to comprehensive fraud signals revealing the context and method of tampering, we provide a singular Authenticity Score, indicating the likelihood that a document is genuine.

The Authenticity Score ranges from 0-100. The score weighs the context of what was tampered with and our confidence in the signal.

We categorize scores as follows:

• 0-55: LOW authenticity
• 56-80: MEDIUM authenticity
• 81-100: HIGH authenticity

Reason codes are returned to offer transparency into the score, and can double as high-level signals that allow for simple logical rules to determine the flow of the document. The score should be a helpful signal to prioritize document flow.

You may find that below or above a certain threshold can be rejected or approved without manual review.

Authenticity Score in the Dashboard

The Authenticity column in the Book List displays the lowest Authenticity Score of any document in a Book. This helps you quickly identify potential fraud. This column is sortable and enables you to prioritize your workflow by focusing on documents with low scores that may indicate severe fraud, or high scores that can be approved with minimal review. Similarly, the Book Overview page shows the lowest score from any document in the upload. The authenticity score is also available on the Detect tab of the document detail page to provide additional information such as reason codes.

The Detect tab in the document details provides an overall authenticity assessment of the document, including the Authenticity Status and Authenticity Score.

The Authenticity Status provides a high-level classification (Low, Medium, High), while the Authenticity Score (0–100) reflects the confidence in that assessment.

Signals that significantly influence the evaluation are highlighted and presented as high-level indicators to support decision-making.

Examples

• LOW

  

• MEDIUM

  

• HIGH

  

Choosing a threshold

The score is derived from an assessment of both the severity of identified signals and the corresponding confidence levels.

As a general guideline, a lower Authenticity Score indicates a higher likelihood of document risk. Scores in the lower range, typically below 60, are more likely to contain strong signals of manipulation or inconsistencies.

However, thresholds may vary depending on your risk tolerance and use case.

For example:

• If your review process is highly risk-averse, you may choose to treat scores below 70 as low authenticity.
• If you are comfortable allowing minor inconsistencies or lower-confidence signals, you may consider only the lowest score ranges such as 30 to 45 and below as low authenticity.

The Authenticity Score should be used as a directional signal. Final decisions should consider the document holistically, along with the specific signals and context available during review.

Authenticity Score in the API

Authenticity Score is returned at the document level. The response includes the numerical score and reason codes.

{
    "form_analysis": [
        {
            "form_type": "PAYSTUB",
            "form_uuid": "37e333e4-bd4b-44fc-bd02-d82bc650140a",
            "signals": [
                {
                    "identifier": "invalid_date",
                    "display_name": "Invalid Date",
                    "signal_count": 1,
                    "page_number": 1,
                    "supporting_data": [
                        {
                            "values": [
                                {
                                    "key": "captured_date",
                                    "value": "2023-05-14",
                                    "data_type": "str"
                                }
                            ]
                        }
                    ]
                }
            ],
            "form_authenticity": {
                "version": "1.0",
                "score": 55,
                "reason_codes": [
                    {
                        "code": "PROCESSED_DOCUMENT",
                        "confidence": "MEDIUM",
                        "description": "Document was resaved using common document tools"
                    }
                ],
                "authenticity_status": "MEDIUM"
            }
        }
    ]
}
        

Each signal includes a unique identifier, a descriptive label of the identifier, and supporting data that provides additional context about the finding. Signals may also include metadata such as occurrence count and page reference where applicable. To learn more, see the Book-level fraud signals and Document-level fraud signals API documentation pages.

In this instance, the system identifies a high-confidence issue related to date validity, along with additional signals indicating how the document was generated or processed. These signals collectively contribute to the overall authenticity assessment.

Using Authenticity Score to optimize your workflow

Dashboard users

In only 3 clicks, you can navigate to visualizations and signals outlining your worst fraud:

1. Sort by Authenticity Score in the dashboard to find high-risk documents that need urgent review.
2. Click on Books with low scores.
3. Then, click on documents within that Book with the lowest scores to review detailed findings.

A similar approach can be used to identify low-risk documents that can be moved forward with minimal review.

A weekly or daily workflow may involve filtering for date range to include documents from the previous day or previous week, then sorting by score to prioritize which documents need review.

API Users

The Authenticity Score can be used to automate workflows based on configurable thresholds. Lower scores generally indicate a higher likelihood of document risk, while higher scores suggest stronger indicators of authenticity.

For example:

• Scores in the lower range such as 30 and below typically correspond to strong signals of manipulation or synthetic content.
• Mid-range scores such as 45 to 60 may indicate processing, editing, or localized inconsistencies that warrant further review.

👍

Tip

Since multiple signals can contribute to the score, it is recommended to use both the score and the underlying signals when making automated or manual decisions.