For agency owners and people operations leads
90-day new-hire cohort retention calculator
For a group of new hires whose first 90 days have all ended, calculate how many remained continuously employed. If a status is unknown, see the range the missing information could change.
The cohort measure
Define a complete cohort
Choose a defined group of people who started employment in the same period, such as new hires starting in one calendar month. Include a person only once in the original group under your documented rules. Choose one consistent way to count each person’s 90-day window from their start date, and use actual employment records to determine whether they remained continuously employed throughout that window. Apply your organization’s documented treatment of transfers, leaves and rehires consistently.
Only calculate when every included person’s full 90-day window has ended. A person whose window is still open is not an unknown completed outcome: wait until the full cohort is observable, or define and report a separate completed cohort with its inclusion rules. Do not classify a rehire after an employment break as continuously employed through the break.
Calculate the cohort result
Enter aggregate counts only. Split the original cohort into exactly three non-overlapping categories: confirmed continuously employed through the full window, confirmed employment ended before completing it, and unknown because the completed outcome cannot be verified. The counts must add to the original cohort.
Worked example
In a completed cohort of 10, 8 are confirmed continuously employed through 90 days, 1 is confirmed to have left before completing 90 days and 1 status is unknown. The confirmed lower bound is 80%; if the unknown person also remained, the upper bound is 90%. Once the unknown status is verified, recalculate. A missing record is not proof that someone left.
Use the result carefully
- Keep the start period, observation rule, inclusion and exclusion rules, and data source beside the result. This cohort measure and a monthly separation rate answer different questions.
- Compare cohorts only with compatible definitions and observation windows. Small groups are sensitive to one person’s status; a rate alone cannot establish a cause or prove a training effect.
- Share only aggregates that cannot reasonably identify someone, including through small group size, dates or combining details. Keep individual records in approved restricted systems.
Keep candidates informed
Use adaptable messages for interview scheduling, an open decision and a selection update.
Open the candidate update templatesExplore the agency training toolkit · More resources for organizations
