Skip to main content




Configuring Predictive Models Best Practices

  • Include as much employee data as possible. It is safe to include all properties that meet your organization's compliance requirements.
  • Include clean data that is free of data artifacts.
  • Exclude any sensitive data such as age or gender.
  • Exclude backdated data.
  • Age and Generation properties can be added at the same time. However, there is no reason to add Generation if Age is already included because there is no additional information provided by Generation.
  • Add all compensation properties as base and variable compensation are distinct pieces of information.
    • Including only Direct Compensation does not reveal the relative size of the individual compensation segments.
    • Include both percentages and absolute amounts for items such as bonuses and long-term incentives and let the predictive model decide what's worth keeping.

For more information about our recommendations, see Strengths and limitations of the predictive models.

  • Was this article helpful?