# MindBridge Overall Score for General Ledger Analyses

Jonathon Plowman-Samson
• Updated

## Summary

Learn how risk scores are calculated for General Ledger Analyses.

## What Makes an Overall Score?

Once MindBridge analyzes your financial data using the control point ensemble, each transaction and entry is given an overall score, depending on the specific control point.

An overall score represents the total risk of the entry or transaction, taking all of the control points into consideration. Control points have their own weights (see more on control point weights), which affects the way each control point impacts the overall score.

Transactions that trigger many control points will likely have a higher overall score. Alternatively, transactions that trigger a single control point that's heavily weighted will also have a higher overall score. Transactions with higher overall scores are more likely to be of interest during audit scenarios.

## Different levels of analysis

Our control points can analyze the following:

• Line items (dollar value)
• Transactions (a collection of line items)
• Flows (the movement of money between accounts - see Understanding Monetary Flow)

### Line items

When analyzing individual line items, the overall score is calculated as follows:

1. The score of each triggered control point is multiplied by the control point’s weight.
2. The individual scores found in Step 1 are summed up into a single score.
3. The sum of the triggered control points found in Step 2 is divided by the total potential weight of all control points (as customized by the user or library default).

### Transactions

The same calculation as above is applied to transactions, but there are some subtle differences. Each line item ‘inherits’ the transaction’s score. If, for example, a transaction triggers the Duplicate control point, each line item in the transaction inherits that Duplicate score.

Control point scores given to individual line items, however, are not promoted to the transaction. Instead, only the highest score for each triggered control point is promoted. If, for example, a transaction contains three line items that trigger Unusual Amount, with scores of 20%, 30%, and 55%, the transaction is given an Unusual Amount score of 55%.

The reasoning behind this method is that, for example, if a transaction has the following:

• A round dollar value
• A high dollar value
• A weekend posting
• Other major anomalies

By choosing this method, we highlight any transaction with multiple anomalies, regardless of if they occur in the same line item or not.

### Flows

For a more in-depth look at monetary flows, see our article, Understanding Monetary Flows

When analyzing a flow, each line item will inherit the flow’s score - just like transaction scores.

There are exceptions to this, and in some cases, single line items may inherit scores from more than one flow.

Transactions with a many-to-one flow (a transaction containing a \$100 debit, with an \$80 and \$20 credit) would be an example of this. The \$100 debit will have two associated flows: one for \$80 and one for \$20. In the case where a line item inherits scores from two or more flows, the line item will inherit the maximum flow score.

## General ledger control point default weights

### 2-Digit Benford

Type: Statistical
Level of Analysis: Entry
Weight: 5%
This control point flags overrepresented digits in a ledger. While it flags misstatement or a control breakdown, it can also flag repetitious behavior that may be normal to business processes. Because of this, it was given a standard weight of 5%.

### Cash Expenditure

Type: Rules-based
Level of Analysis: Entry
Weight: 10%
Domain expert opinions found that cash expenditures are by nature a high-risk business area. Because of this, the control point received an elevated weight of 10%.

### Cash to Bad Debt Conversion

Type: Rules-based
Level of Analysis: Transaction
Weight: 20%
This control point checks very strict conditions across multiple transactions. Therefore, it's a strong indicator of misstatement. Since it's such a strong indicator, it was given a weight of 20%.

### Complex Instrument

Type: Rules-based
Level of Analysis: Entry
Weight: 1%
Based on the number of instances where this control point is seen to be occurring, it was seen as minimally impactful. Therefore, it was given a minimal weighting of 1%.

### Complex Structure

Type: Statistical
Level of Analysis: Transaction
Weight: 1%
Based on the number of instances where this control point is seen to be occurring, it was seen as minimally impactful. Therefore, it was given a minimal weighting of 1%.

### Duplicate Transactions

Type: Rules-based
Level of Analysis: Transaction
Weight: 5%
This control point was identified by domain experts as being related to misstatement, but not to the extent that its weight should be elevated. So, it was given the standard weight of 5%.

### Empty Text Field

Type: Rules-based
Level of Analysis: Entry
Weight: 1%
Domain expert opinions found that this control point does not indicate misstatement by itself. But when combined with other control points, it can help to triangulate misstatement. Therefore, it was given a minimal weight of 1%.

### End of Reporting Period

Type: Rules-based
Level of Analysis: Entry
Weight: 1%
Domain expert opinions found that this control point does not indicate misstatement by itself. But when combined with other control points, it can help to triangulate misstatement. Therefore, it was given a minimal weight of 1%.

### End of Analysis Period

Type: Rules-based
Level of Analysis: Entry
Weight: 1%
Domain expert opinions found that this control point does not indicate misstatement by itself. But when combined with other control points, it can help to triangulate misstatement. Therefore, it was given a minimal weight of 1%.

### Expert Score

Type: Machine learning
Level of Analysis: Transactional
Weight: 10%
Our testing has shown that this control point correlates with misstatement. Therefore, it received an elevated weight of 10%. It also factors into the Flow Analysis control point, increasing its effect on the overall score.

### Flow Analysis

Type: Machine learning
Level of Analysis: Transactional
Weight: 80%
This control point combines our machine learning algorithm scores, which has proven to be a very strong indicator of misstatement and anomalous transactions. We have conducted several studies internally, as well as with partners and customers, and determined that a default weight of 80% is appropriate.

### High Monetary Value

Type: Rules-based
Level of Analysis: Entry
Weight: 10%
This control point received an elevated score of 10%, owing to the fact that misstatements within transactions of high monetary value have the largest effect on a business’s financial position.

### Last 3 Digits

Type: Rules-based
Level of Analysis: Entry
Weight: 5%
Domain expert opinions found that this control point is related to misstatement, but not warranting special attention. Therefore, it was given the standard weight of 5%.

### Manual Entry

Type: Rules-based
Level of Analysis: Entry
Weight: 10%
Within the context of a general ledger, domain expert opinions found that entries entered manually correlate with misstatement by nature. Therefore, it was given an elevated weight of 10%.

### Material Value

Type: Rules-based
Level of Analysis: Entry
Weight: 0%
This control point is set to 0% by default since it requires auditor input. Your organization needs to determine the material value threshold (dollar value) and weight. Any transactions exceeding your threshold will trigger this control point.

Depending on the importance you place on material value in your audits, you can weigh this control point accordingly. We recommend that your weight does not exceed our highest default weight of 10%.

### Outlier Anomaly

Type: Machine learning
Level of Analysis: Transactional
Weight: 5%
Based on the impact of including machine learning, this control point was seen as moderately impactful. Therefore, a standard weighting of 5% was given.

### Rare Flow

Type: Machine learning
Level of Analysis: Transactional
Weight: 10%
Our testing has shown that this control point correlates with misstatement. Therefore, it received an elevated weight of 10%. It also factors into the Flow Analysis control point, increasing its effect on the overall score.

### Reporting Period Adjustment

Type: Rules-based
Level of Analysis: Transactional
Weight: 1%
This control point fires on very strict conditions and based on the number of instances where this control point is seen to be occurring it was seen as minimally impactful. Therefore, it was given a minimal weighting of 1%.

### Reversal

Type: Rules-based
Level of Analysis: Transactional
Weight: 5%
Domain expert opinions found that this control point is related to misstatement, but not warranting special attention. Therefore, it was given the standard weight of 5%.

### Reversed

Type: Rules-based
Level of Analysis: Transactional
Weight: 5%
Domain expert opinions found that this control point is related to misstatement, but not warranting special attention. Therefore, it was given the standard weight of 5%.

### Sequence Gap

Type: Rules-based
Level of Analysis: Transactional
Weight: 1%
This point identifies transactions that have been removed from a general ledger, but it triggers the transactions that occur before and after the missing transaction.

Since it does not flag the missing transaction, we do not want this control point to have an overly aggressive impact on the scores of the neighboring transactions.

### Start of Reporting Period

Type: Rules-based
Level of Analysis: Transactional
Weight: 1%
Domain expert opinions found that this control point does not indicate misstatement by itself. But when combined with other control points, it can help to triangulate misstatement. Therefore, it was given a minimal weight of 1%.

### Start of Analysis Period

Type: Rules-based
Level of Analysis: Entry
Weight: 1%
Domain expert opinions found that this control point does not indicate misstatement by itself. But when combined with other control points, it can help to triangulate misstatement. Therefore, it was given a minimal weight of 1%.

### Suspicious Keyword

Type: Rules-based
Level of Analysis: Entry
Weight: 5%
Domain expert opinions found that this control point is related to misstatement, but not warranting special attention. Therefore, it was given the standard weight of 5%.

### Unbalanced Debits & Credits

Type: Rules-based
Level of Analysis: Transactional
Weight: 5%
Domain expert opinions found that this control point is related to misstatement, but not warranting special attention. Therefore, it was given the standard weight of 5%.

### Unusual Amounts

Type: Machine learning
Level of Analysis: Entry
Weight: 5%
Based on the impact of including machine learning, this control point was seen as moderately impactful. Therefore, a standard weighting of 5% was given.

### Weekend Post

Type: Rules-based
Level of Analysis: Entry
Weight: 5%
Domain expert opinions found that this control point is related to misstatement, but not warranting special attention. Therefore, it was given the standard weight of 5%.

### Zero Entry

Type: Rules-based
Level of Analysis: Entry
Weight: 5%
Domain expert opinions found that this control point is related to misstatement, but not warranting special attention. Therefore, it was given the standard weight of 5%.

## Not-for-profit general ledger control point default weights

### Expense Flurry

Type: Machine learning
Level of Analysis: Transaction
Weight: 10%
This control point received an elevated weight of 10% in not-for-profit libraries since it detects scenarios that have been determined by domain experts as being highly relevant to not-for-profit audits.

### Fund Expense Flurry

Type: Machine learning
Level of Analysis: Transaction
Weight: 10%
This control point received an elevated weight of 10% in not-for-profit libraries since it detects scenarios that have been determined by domain experts as being highly relevant to not-for-profit audits.

### Interfund Transfer

Type: Rules-based
Level of Analysis: Transaction
Weight: 10%
This control point received an elevated weight of 10% in not-for-profit with fund libraries since it detects scenarios that were determined by domain experts as being highly relevant to not-for-profit audits.

### Split Expense (Multiple Transactions)

Type: Rules-based
Level of Analysis: Transaction
Weight: 10%
This control point received an elevated weight of 10% in not-for-profit libraries since it detects scenarios that were determined by domain experts as being highly relevant to not-for-profit audits.

### Split Expense (Single Transaction)

Type: Rules-based
Level of Analysis: Transaction
Weight: 10%
This control point received an elevated weight of 10% in not-for-profit libraries since it detects scenarios that were determined by domain experts as being highly relevant to not-for-profit audits.

Anything else on your mind?
Log in to MindBridge to chat with us, submit a request, or reach out to your assigned Customer Success Manager.