Employee at Risk in Talenta Insights displays Predictive Attrition Risk, which is a prediction of the Employee Attrition rate. Employee attrition (a reduction in the number of employees) is a condition in which employees gradually leave the company. This can occur due to resignation, age, health, and other factors. This prediction applies only to the next month.
Example: If you open the dashboard in December 2026, the system will calculate data from November 1–31, 2026, to make a prediction for January 2027.

The following is an explanation of the data presented on the Employee at Risk page:

• Accuracy Score (Stat Card - %): The prediction model's accuracy percentage in predicting whether employees resigned or stayed in the previous month (based on the Critical Risk indicator).

• View Factors (Info Panel): An explanation panel for the 16 risk factors analyzed by AI (such as length of service, salary increases, overtime, time off, and job level), along with their influence on resignation risk.

• Summary Predictive Attrition Risk (Stat Cards): The company's average resignation risk score, along with the number of employees divided into 4 risk levels:

  • Low (0% – 25%)
  • Medium (25% – 50%)
  • High (50% – 75%)
  • Critical (75% – 100%)

• Overview Attrition Risk (Pie Chart): The comparative proportion of employees at each risk level for the following month.

• Attrition Risk Trends (Line Chart with Alert): Month-to-month risk trends accompanied by color-coded alerts:

  • Yellow: Predicted increase in the number of leavers (employees leaving).
  • Green: Predicted decrease in the number of leavers.
  • White: The change tends to be stable / there is no significant change.

  • Attrition Risk Breakdown (Bar Chart): The distribution of resignation risk levels based on Organization, Branch, Gender, and similar variables.

• Risk vs Protective Factor (Bar Chart): The top 10 factors that most increase resignation risk (risk factors) versus factors that reduce or prevent potential resignations (protective factors).

• Employee Attrition Details (Data Table): A detailed individual-level table displaying each employee's resignation risk score along with their complete profile.

Important
The factors used as the basis for the analysis will differ between companies and will change over time. Some of the main common factors used are length of service, average schedule, employment status, 1-year/6-month salary increases, time off, overtime, average actual working hours, and division.

Follow these steps to access information for Employee at Risk.

1. Go to insights.talenta.co, then Sign in to your account.
2. After entering the Talenta Insights page, click the "Employee At Risk" menu.
   
3. You will be taken to the Employee at Risk page.
4. If you want your Employee at risk dashboard to display a specific date range and/or a specific branch/organization, set the filter fields, then click “Apply filter”.
   

5. In the upper-right section of the page, you can see the following display.
   

   No.	Column/Button Name	Description
   1	Accuracy score	An accuracy indicator (in percentage form) showing how accurate the modeling used by Talenta is. The system calculates the total number of predicted resignations that actually occurred and the total number of predicted non-resignations that correctly did not occur. The value displayed differs for each company depending on the number of employees and the number of employees who use ESS (Employee Self Service) Talenta. The more employees and employees who use ESS Talenta, the more accurate the value. For more information, see the Important box below the table.
   2	View factors	Click to open an explanation of the key factors.   The factors below are determined using InterpretML. With InterpretML, you can see the factors that most influence the prediction of employee resignations. In addition, this feature uses LightGBM to predict employee resignations. The following is an explanation: Working period Assesses the total duration of active employment, measured from the start date. Employees who stay longer are less likely to resign. Average schedule Calculates the average number of hours an employee was scheduled to work during the last month. Fewer working hours mean the employee is likely to stay longer. Employment status Analyzes employment status (intern, contract, permanent) in relation to resignation risk. Permanent employees are less likely to resign. 1 year's salary increment Measures the salary increase over one year. A higher increase correlates with lower resignation risk. Time off Tracks the total time off taken by an employee in a month. Employees who take more time off are more likely to resign. Overtime Tracks how often an employee works overtime in a month. Less overtime means the employee is likely to work longer. 6 months' salary increment Measures the salary increase over the last 6 months. A higher increase correlates with lower resignation risk. Average actual work The average actual working hours of an employee in a month. The lower the actual working hours, the lower the employee's likelihood of resigning. Organization Analyzes turnover rates by department to identify areas with a high tendency for turnover. Lower turnover often indicates effective management practices. 2 years' salary increment Measures the salary increase over the last 2 years. A higher increase correlates with lower resignation risk. Job level Reviews various job levels from low to high. Employees at higher job levels tend to be less likely to resign. Age Analyzes resignation trends across age groups. Older employees generally have lower resignation rates. Late in Tracks the total number of days an employee is late each month. Frequent lateness may indicate a higher resignation risk. Tax status Analyzes employee dependents for PTKP (Taxable Income) purposes. Employees with more dependents often stay employed longer. Average early out Tracks the average time an employee leaves early each month and the number of days this occurs. Frequently leaving early may indicate a higher resignation risk. Reprimand Counts the number of reprimands (warning letters) an employee receives in a month. Employees who receive fewer reprimands tend to be less likely to resign. Click the “X” icon to close the pop-up.
   3	Help	Click to open the following dropdown. - Guidebook: Click to open this Employee at Risk guide article. - Quick tour: Click to get a brief guided tour directly in Talenta. - Share feedback: Click to open a pop-up where you can provide feedback and suggestions about this feature.

Important
Factors Affecting Accuracy in the Accuracy Score:
- Accuracy varies between companies, depending on the number of employees and use of ESS Talenta.
- The more employees and ESS Talenta users there are, the more accurate the prediction.
To evaluate the effectiveness of the prediction model, this feature uses Precision and Recall.
- Precision: This metric assesses how accurate the predictions made by the model are. In the context of this dashboard, precision answers the question: "Of all the employees the prediction model says will resign, how many will actually resign?"
High precision means that when the model predicts an employee will resign, the prediction is likely to be correct.
Example: If the model predicts that 100 employees will resign and 80 of them actually resign, then precision is 80%.
- Recall: This metric assesses the model's ability to identify all relevant cases. In the context of this dashboard, recall answers the question: "Of all the employees who actually resign, how many did the model successfully predict would resign?" High recall means the model is effective at identifying most employees who actually resign.
Example: If 100 employees actually resign and the model successfully predicts 70 of them, recall is 70%.
Both metrics are important because they provide a balanced view of the model's performance. High precision with low recall may mean the model is too cautious and predicts resignation only when it is very certain, but misses many other cases. Conversely, high recall with low precision may mean the model predicts resignation too often, resulting in many incorrect predictions.

The following is an explanation of each piece of information presented on the Employee at Risk page.

A. Summary Predictive Attrition Risk

In this first section, you can see a brief category-based summary of the Attrition Risk prediction, displayed as percentages and the number of employees.

This summary information is grouped as follows:

B. Overview Attrition Risk

In this second section, overall information is presented regarding the prediction of the likelihood that employees will resign or leave the company in the following month. The information is grouped by level:

• Low risk, the likelihood of an employee resigning is below 25%.
• Medium risk, the likelihood of an employee resigning is between 25% and 50%.
• High risk, the likelihood of an employee resigning is 50%–75%.
• Critical risk, the likelihood of an employee resigning is 75%–100%.

You can download the chart as a spreadsheet by clicking “Download”, which will direct you to the Export history page.

C. Attrition Risk Trends

In this third section, you can see a brief category-based summary of the Attrition Risk prediction, displayed as percentages and the number of employees.

 

- Show percentage number: Check this option if you want the chart to display the Attrition Risk Trends percentage. Click it again to hide the percentage.
- Download: Click this button to download the chart as a spreadsheet.
- Notification Alert bar:
Yellow means the system predicts an increase in the percentage of employees leaving the company compared with the previous month.

Green means the system predicts a decrease in the percentage of employees leaving the company compared with the previous month.

White means the system predicts that the percentage of employees leaving the company will remain the same compared with the previous month.

D. Attrition Risk Breakdown

In this fourth section, predictions are presented based on employee grouping filters, such as Organization, Branch, Gender, and others. You can view more detailed information on the right side of the screen.

 

- Search bar: You can enter a specific keyword to search for information in this chart section.
- Download: Click this button to download the chart as a spreadsheet.
- Breakdown by: Adjust the chart display according to your needs. The example in the image above is broken down by Organization. If you select Breakdown by Organization (and certain other breakdown categories), you can add details under View by.

E. Risk vs Protective Factor

In this fifth section, you can see the 10 (ten) factors that most influence whether employees resign (Risk factor) or stay (Protective factor). This allows the company to identify which factors need to be improved to retain its employees.

 

- You can download the chart as a spreadsheet by clicking “Download”.
- You can hover over one of the bars to see its details. A red factor with a positive Risk score value (moving to the right) means that the greater the factor, the higher the probability that an employee will resign. The percentage shown is compared with the previous month. The following is a breakdown of the Risk factor (red).

The following is a breakdown of the Protective factors (green). A green factor with a negative Risk score value (moving to the left) means that the greater the factor, the lower the probability that an employee will resign. The percentage shown is compared with the previous month.

F. Employee Attrition Details

In this sixth section, you can see a list of employees, along with their complete personal information and resignation likelihood level.

That concludes the overview of Employee at Risk. Next, you can learn about Employee Performance here.