Fraud Radar is a Dashboard page that brings your fraud results together in one place. It shows how many applicants you reject for fraud and how this number changes over time, which applicant groups have more fraud than others, and examples of individual applicants and fraud networks.

Fraud Radar helps you answer three questions:

• What happens — how the number of fraud cases and applicant risk levels change over time.
• Where it happens — which countries, document types, platforms, and other groups have more fraud.
• Who is behind it — which applicants and fraud networks are connected to the problem.

Use Fraud Radar to:

• Get the full picture before you review individual applicants.
• Identify the applicant groups that need more attention and decide where your team starts its review.
• Go from a number on a chart to the list of applicants behind it, without exporting data or building your own reports.
• Make sure that your applicant risk levels match confirmed fraud before you use them for automated decisions.

Fraud Radar uses data that Sumsub already collects during applicant checks. You do not need to do any technical setup.

Prerequisites

Fraud Radar is available to all clients. The page sections you see depend on the Sumsub services you use and the permissions of your Dashboard user role.

Get started with Fraud Radar

In the Dashboard, go to Monitoring → Fraud radar.

Use the filters at the top of the page to select a period and the applicants you want to analyze. All page sections use the same filters. For example, if you select one country, all charts, tables, and examples show only applicants from that country.

Fraud Radar metrics

Fraud Radar counts only reviewed applicants. Each applicant is counted once, based on the latest review decision.

Fraud Radar treats the following rejection labels as fraud:

• Forgery
• Fraudulent liveness
• Fraudulent Patterns
• Selfie mismatch
• Bad face matching
• Blocklist
• Suspected stolen documents
• Third party involvement

Fraud Radar sections

The page moves from the general picture to individual applicants: totals and trends first, then applicant groups, then examples of applicants and fraud networks.

Reviewed applicants

The Reviewed applicants chart shows the number of reviewed applicants and how the Fraud rate and High-risk profile rate change over the selected period.

• Bars show the number of reviewed applicants, measured on the right scale, Applicants.
• Lines show the two rates, measured on the left scale, Rate.

Totals for the full period are displayed above the chart.

Risk levels

The Risk levels chart shows how many applicants received each risk level in the selected period. Level names and colors match your applicant risk scoring settings. Applicants without a risk level are grouped under (no risk level).

Use this chart to track increases in high-risk applicants and to check that levels are distributed as you expect. To change how Sumsub assigns risk levels, select Manage risk scoring.

If you have no risk levels configured, the chart shows the No risk levels configured yet message. To add risk levels, select Set up. For more information, refer to Fraud Prevention Starter Kit.

Applicants breakdown

The Applicants breakdown table splits reviewed applicants into groups by one property at a time. Use it to find groups with more fraud than others. To select the property, open one of the following tabs:

• Reject label
• Risk level
• Risk label
• Document country
• Document type
• IP country
• Level
• Source key
• Tag
• Platform

For each group, the table shows Reviewed applicants, Rejected for fraud, Fraud rate, Rejected for high-risk profile, and High-risk profile rate.

• Share of total. Hover over a number of applicants to see its share of the corresponding total in the Reviewed applicants chart, with the same filters applied. For example, if the chart shows 200 applicants rejected for fraud and one country has 50, hovering over 50 shows 25% of applicants rejected for fraud.
• Difference from average. Below each rate, the table shows how much the group rate differs from the average. The average is the rate in the Reviewed applicants chart for all applicants matching your filters. For example, if the average Fraud rate is 2% and one country has 2.7%, the table shows +35% vs avg.

Applicants examples

The Applicants examples section shows example applicants matching your current filters. Use it to review the decisions behind the numbers. Select one of the following tabs:

• Approved
• Rejected for fraud
• Rejected for high-risk profile

Each tab shows the applicants with the highest and the lowest risk scores. Compare the decision with the risk level of each applicant to interpret the results.

If you have no risk levels configured, this section shows the newest and the oldest applicants, together with a message on how to add risk levels.

Largest applicants networks examples

The Largest applicants networks examples section shows the largest fraud networks that include applicants matching your current filters.

A fraud network is a group of applicants connected by the same data, for example, the same device. Each card shows the number of applicants, the network ID, the network status, and the pattern detected by Sumsub.

To open a network, select its card. To see all networks, select View all.

Investigate fraud increase

To go from a number on a chart to the applicants behind it:

1. Find the increase. In Reviewed applicants, look for a sudden increase in the Fraud rate or High-risk profile rate. Set the Created filter to the days around the increase.
2. Find the group. In Applicants breakdown, open the tabs one by one and sort by Fraud rate. Look for groups with a high rate and a red vs avg. value, for example, a single country, source key, or platform.
3. Filter by the group. Add the group as a filter. All page sections now show only this group.
4. Review the applicants. In Applicants examples, review the applicants with the highest risk scores, especially on the Approved tab. To see all applicants, select View full table, or select a number in the breakdown table.
5. Check for connections. In Largest applicants networks examples, check whether these applicants are connected through fraud networks.

After you identify the cause, use other Sumsub tools to address it:

• Adjust your applicant risk scoring rules so that these applicants receive a higher risk level.
• In Workflow Automation, route these applicants to additional checks or to manual review.
• Add confirmed fraudsters to the​ blocklist. This helps Sumsub detect connected applicants in the future.

Validate risk levels

Before you use risk levels for automated decisions, make sure that they match confirmed fraud:

1. In Applicants breakdown, open the Risk level tab.
2. Compare the Fraud rate across levels. Higher risk levels must have a higher fraud rate than lower ones.
3. In Applicants examples, open the Approved tab and review the approved applicants with the highest risk scores.

If higher risk levels do not have a higher fraud rate, or if many approved applicants have high risk levels, adjust your risk scoring rules. For more information, refer to this article.

Improve data accuracy

Fraud Radar shows only the data that Sumsub has. The more risk signals and confirmed fraud cases Sumsub receives, the more complete and accurate the page becomes.

Use Fraud Prevention Starter Kit

Fraud Prevention Starter Kit provides a full set of risk signals and a fraud risk level for each applicant, without building your own rules. With the Starter Kit, you get:

• Risk signals from email, phone, IP address, device, selfie, and connections between applicants.
• A risk level for each applicant, for example, Low, Medium, or High.
• The Risk levels chart, the Risk level tab and filter, and the Applicants examples section.

To add the Starter Kit, contact your Customer Success Manager at [email protected].

Report fraud detected outside Sumsub

You can detect fraud after verification, for example, from external reports, chargebacks, or user activity after onboarding. Report each of these applicants to Sumsub.

Each reported applicant makes your statistics more accurate:

• The Fraud rate and the breakdown table show the full amount of fraud, not only fraud detected during verification.
• You see which countries, documents, and platforms bring fraud that verification did not stop.
• Sumsub uses this information to detect connected applicants in the future and to assign more accurate risk levels.