Preventing employee turnover is crucial for any organization, and data can play a pivotal role in highlighting the importance of retention. By leveraging data to understand the costs, productivity, engagement, and impact on company reputation, organizations can make informed decisions about strategies to retain talent. Preventing turnover isn't just about keeping employees; it's about sustaining long-term growth, maintaining operational efficiency, and protecting both financial and intangible resources.
1 What information Nailted gathers for forecasting turnover?
Nailted’s prediction algorithms uses machine learning to predict the likelihood of an employee leaving the company, offering an advanced tool for HR and management teams to proactively manage employee retention applying the required retention strategies.
The model is based on historical employee data, including both current employees and those who have left the company. Nailted forecasting turnover model uses a variety of demographic, professional and engagement-related variables to make its predictions.The type of variables used, which may include but are not limited to are the following:
- Individual demographic information: Personal information about each employee that helps to understand the composition and diversity of the different employee groups is used to forecast employee turnover.
💡In order to keep your data up to date, we recommend you to integrate your HR software with Nailted. See here the full list of supported HR softwares.
-
Individual Professional Information: This type of attributes refer to the employee journey within the organization.
- Tenure of the person in the company: Amount of time that an employee has been working at the company.
- Turnover: The information used is the aggregate turnover of all the groups that the individual is part of.
-
Aggregated Team and group metrics: This information is gathered from the surveys that Nailted sends to employees.
- Participation of the groups of the person
- eNPS of the groups of the person
- Engagement of the groups of the person
💡Nailted forecasting turnover model will never use individual participation, eNPS or engagement responses. The average of the group's values in which the employee is included will be used to protect individual privacy and prevent potential biases.
Consult further information about Nailted’s turnover forecasting model here.
2 What is turnover calibration?
Turnover calibration refers to the process by which the Nailted turnover forecasting model refines its predictions based on the volume and quality of data it has accumulated on employee turnover rates. This data is drawn from analyzing the number of employees, their departures, and a variety of employee attributes that help identify patterns associated with turnover.
As more data is gathered, the model becomes better at distinguishing between different turnover scenarios, allowing for more accurate forecasting. This calibration process ensures that the model's predictions become increasingly reliable over time, as it learns from historical trends and contextual factors such as tenure, job role, engagement, etc. The higher the volume and diversity of data, the more precisely the model can detect subtle signals that contribute to employee departure, leading to better-prepared strategies for retention.
There are three different calibration ranges:
-
Calibration between 0 - 50%
Nailted turnover forecasting model doesn’t have enough information to provide a forecast. An organization can be on this calibration range due to one or both of the following causes:
-
- Missing employee attributes: You haven’t inserted the required attributes to predict employee turnover. See here the attributes that Nailted supports and add as many as possible.
- Lack of clear patterns: Although the model has enough information from employees attributes, the system cannot detect clear patterns within the departures and the data extracted from attributes to offer a reliable turnover forecast.
💡Turnover calibration may be very low when Nailted is implemented within an organization as the model doesn’t have enough historical data to find clear patterns.
-
Calibration between 51 - 80%
The first turnover predictions can be displayed although the calibration of the model can be improved. Make sure that all the employee attributes and leaving reasons are correctly added.
-
Calibration higher than 81%
The forecasting model has a high calibration percentage, therefore the forecasting is accurate according to the data gathered by the model. It is very important to keep the calibration percentage the highest as possible to guarantee the greatest accuracy in the predictive data.
3 How can I improve my calibration level?
Nailted turnover forecasting models require data to offer predictions, so it is key that the information is added into the system. To improve your calibration level you can do the following:
- Review that each employee has correctly adjusted all their attributes. See here the list of available attributes and how to add them into Nailted.
- Review that each leave has correctly set their leaving reason. This information will be crucial in order to gather the correct employee data and locate patterns to predict turnover. Learn more about how to get accurate turnover data.