Overview

In addition to generating a risk score, Begini evaluates the reliability of each assessment session.

This is captured through trust or confidence indicators, which help identify whether the behavioural data is genuine, consistent and suitable for decisioning.

This layer is critical for detecting manipulation, low-quality inputs and potentially fraudulent behaviour.

---

Why trust matters

Not all completed assessments represent genuine or reliable behaviour.

Without a way to assess data quality, lenders risk:

• Making decisions based on unreliable inputs
• Missing signs of manipulation or assisted completion
• Reducing overall model performance
• Introducing unnecessary risk into the portfolio

The Trust Score helps ensure that results are interpreted in the correct context.

---

What the Trust Score represents

The Trust Score reflects the quality and reliability of the behavioural data captured during the assessment.

It is not a measure of credit risk.

Instead, it answers:

---

What influences the Trust Score

The Trust Score is derived from multiple behavioural signals observed during the session.

Consistency of behaviour

• Stable patterns across tasks
• Logical progression of decisions
• Low contradiction between actions

Inconsistent behaviour may indicate low engagement or manipulation.

---

Response patterns

• Extremely fast responses
• Uniform or repetitive answer patterns
• Lack of variation across tasks

These can indicate random answering or scripted behaviour.

---

Engagement quality

• Completion behaviour
• Drop-offs or interruptions
• Signs of disengagement

Low engagement can reduce the reliability of results.

---

Repeated attempts

• Multiple sessions from the same user
• Attempts to improve outcomes through repetition

This may indicate gaming of the system.

---

Assisted or coached behaviour

• Patterns that suggest external help
• Behaviour inconsistent with natural interaction

This can reduce confidence in the authenticity of the session.

---

How to use the Trust Score

The Trust Score should be used alongside the risk score to guide decisioning.

High trust

• Behaviour is consistent and reliable
• Results can be used with confidence
• Suitable for automated or standard decisioning

---

Medium trust

• Some variability or minor concerns
• Results can still be used, with awareness
• May benefit from additional checks

---

Low trust

• Behaviour is inconsistent or suspicious
• Results should be treated with caution
• Consider manual review or re-assessment

---

Combining risk and trust

The most effective use of Begini comes from combining:

• Risk Score → likelihood of default
• Trust Score → reliability of the data

This allows you to:

• Approve high-quality, low-risk applicants confidently
• Flag low-trust sessions for review
• Avoid making decisions based on unreliable inputs

---

Detecting fraud and manipulation

Begini’s behavioural approach allows it to identify patterns that are difficult to detect through traditional methods.

This includes:

• Randomised answering
• Automated or scripted behaviour
• Repeated attempts to influence outcomes
• Assisted completion

These signals are captured as part of the assessment process without requiring additional user input.

---

Best practices

• Always consider Trust Score alongside Risk Score
• Define internal rules for handling low-trust sessions
• Monitor patterns of repeated or suspicious behaviour
• Use Trust Score to support fraud and risk workflows
• Avoid fully automated decisions on low-trust results

---

Common mistakes to avoid

• Ignoring trust indicators and relying only on risk score
• Treating all completed assessments as equally reliable
• Not defining a process for low-trust sessions
• Overriding trust signals without review

---

Next steps

To complete your understanding of the assessment: