Review flight risk

What this achieves

Flight risk estimates how likely each employee is to leave, from patterns already in your data, and lists the people scored highest. It exists to start conversations early, not to reach conclusions about anyone.

Required role: Administrator.

A score is a prompt for a conversation

Read this before you read any number on the screen.

A flight-risk score is a model’s estimate from patterns in your own records. It is not a statement about what a person intends to do, and it cannot be. Someone can score high and be perfectly settled; someone can score low and resign the following week.

Treat a high score as a reason for a manager to have a conversation, and treat the indicators behind it as more informative than the score itself. Any decision that follows is a person’s decision, made with the employee involved.

Example: a high score for James Whitfield is a prompt for Olivia Bennett to talk to him about his development plan and what he wants next. It is not a reason to change his work, his pay, or his prospects, and he should not learn about it as a fact recorded against him.

Steps

Go to Performance → Insights → Flight Risk.
Select Recalculate All to generate scores across active employees.
Optionally narrow the view with the Department filter.
Read the counts, then the High & critical risk employees list.
Select an employee in the list to open their risk detail.
Read the Risk factors, then select Acknowledge.

What the figures count

Option	Description
Critical, High, Medium, Low	Employees at each risk level. Only employees who have a score are counted.
Total assessed	How many employees have a score at all.
Avg score	The average score across assessed employees.

Total assessed is the figure that gives the others meaning. Everything on this screen describes the assessed population, so a low high-risk count on a small assessed population says very little.

Example: if 60 people at HC Corp UK Ltd have been scored and Engineering has 34 employees, a count of 4 high-risk in Engineering only makes sense once you know how many of those 34 were scored.

Scores are as at the moment they were calculated

Each score carries the date it was calculated, shown on the employee’s risk detail. Nothing recalculates on its own when you open the screen.

That means a score can be stale. If your last calculation was months ago, the list reflects the company as it was then. Recalculate before acting on it.

Recalculate All covers active employees in one run and takes time to complete. Recalculate inside an employee’s risk detail refreshes one person.

Reading an employee’s risk detail

Option	Description
Risk score	The score and its level, with the date it was calculated.
AI analysis	A written explanation of the score, where one has been generated.
Risk factors	The individual indicators that contributed, each with the points it added.
Suggested actions	Suggested next steps, where the analysis produced any.

The Risk factors list is the useful part. It tells you what the score is actually built from, which is what a manager needs before any conversation — a score on its own gives them nothing to talk about.

Example: reading that tenure and time since last promotion are the two largest contributors for James Whitfield tells Olivia Bennett what the conversation should be about. The number 78 does not.

Acknowledging a score

Acknowledge records that somebody looked at this employee’s risk. It stores who acknowledged it and when, and it stops the same score being rediscovered as if it were new.

Acknowledging is not a resolution and it does not change the score. It is a record of attention, which matters when several people are looking at the same list.

Risk distribution and departments

The distribution chart shows the same counts as a proportion, with Total assessed in the centre. Beneath it, Top departments by high risk lists departments as high-risk over total.

Be careful reading department figures for small departments. Finance has 9 people at HC Corp and People has 6, so one or two high-risk employees is a large proportion in either — and a proportion small enough to identify individuals is not a statistic, it is a person.

What happens next

The list is the start of manager conversations, not the end of an analysis. Where several high scores share the same indicators, that pattern is worth more than any individual score and usually points at something structural — a team, a grade, or a length of service. Where an employee’s factors point at a development gap, a development plan or a career conversation is the response the product supports.