Generate a workforce prediction

What this achieves

A prediction projects forward from what your records already show — where skill gaps are widening, how much attrition to expect, who is coming through, and how many people you will need to hire. Each one is generated when you ask for it and describes the moment it was generated.

Required role: Administrator.

Steps

Go to Administration → AI Governance → Predictions.
Select the tab for the prediction you want.
Select Generate Prediction.
Choose the Prediction type, the Scope, and the Horizon.
Select Generate.

The five predictions

Option	Description
Skill Gaps (6mo)	Projected skill shortfalls six months out.
Skill Gaps (12mo)	The same analysis at twelve months.
Attrition	Employees by risk level, projected attrition, and the skills concentrated in at-risk employees.
Promotion Pipeline	People by readiness category now, and how many are projected to be ready at the horizon.
Hiring Needs	Current vacancies set against projected attrition and projected internal fills, giving a net hiring need.

Each tab shows only the most recent prediction of that type. Generating again replaces what you are reading; it does not add a second entry alongside it.

Example: generating a twelve-month attrition prediction for HC Corp in April and again in July leaves July’s on screen. If you need April’s figures for a comparison, keep them outside the product or take a workforce snapshot instead.

Choosing the scope and horizon

Option	Description
Scope	Company, Department, or Team. Departments and teams require you to pick which one.
Horizon	3 months, 6 months, or 12 months. How far ahead the projection runs.

A longer horizon is not a better prediction. It projects the same present-day data further, so the further out you look the more the result depends on nothing changing.

Example: a twelve-month hiring-need projection for Engineering at HC Corp assumes Engineering keeps behaving as it does today across 34 people. A departmental reorganisation invalidates it the day it happens.

Read the badges before the numbers

Every generated prediction carries a row of badges: the date it was generated, its horizon, and a confidence percentage.

Confidence is the prediction’s own statement of how much weight it will carry. A low confidence figure is the product telling you the underlying data is thin, and it is a reason to fix the data rather than to act on the projection.

Where the prediction lists Assumptions, read them. They state what the projection took as given, and they are the fastest way to see whether it applies to your situation.

Figures are aggregated, the narrative is written

The numbers come from aggregating what is already recorded — employee skills, readiness, flight risk, and positions. They are arithmetic over your own data, not an opinion.

The narrative alongside them is written text summarising those figures. Treat it as a summary you can check against the tables, not as a separate source of information.

Predictions about people are prompts, not verdicts

The attrition and promotion-pipeline predictions describe groups of people, and the attrition tab links through to the individuals behind it.

Where a prediction names or reaches an individual, it is a reason for a conversation and a decision a person makes with them. Nobody’s prospects should turn on a projection they were never told about.

Example: a promotion-pipeline projection showing four people ready within a year at HC Corp is a prompt for four career conversations. It is not a shortlist, and it is not a decision.

Small groups identify people

Scope a prediction to a department and the group gets small quickly. Finance has 9 people at HC Corp and People has 6, so a departmental attrition projection describing “one or two” is describing individuals whether or not it names them.

Keep small-group output inside the group of people who would legitimately handle it as personal information.

What happens next

A prediction is an input to planning, not a plan. A skill-gap projection becomes learning or recruitment; a hiring need becomes requisitions; an attrition projection becomes retention work owned by managers. Where you want the product to carry the follow-up, generate an action plan from the prediction so the remediation items are tracked rather than remembered.