Do promotions follow the ratings?

The talent process makes a promise: rate well and you advance. Of 42 promotions in twelve months, 30 landed on the top two bands and 29 came from the high-potential column. The promise mostly holds, and the exceptions now each owe one sentence.

RuleTest the promise: promotions should land on high ratings and high potential. Count where they actually landed, every cycle, and explain the exceptions.

Neveraudit only who was promoted. Who was rated ready and passed over is the half of the story that predicts resignations.

◆ Twelve months of promotions, by ratingthe 42, against the rating that preceded them.

Rating at cycle close	Cohort	Promoted in 12mo
5, exceptional	18	8
4, exceeds	92	22
3, meets	151	12
2, developing	29	0
1, below	6	0
Rated cohort	296	42

These are the same 42 promotions the outcomes page counts, re-cut against the rating that preceded them. Thirty of 42 went to the top two bands: the process mostly keeps its promise. The twelve threes each deserve one sentence of why. Sample values are illustrative, never client data.

◆ The same 42, by nine-box columnpotential called it better than performance.

Potential at cycle close	Workers	Promoted in 12mo
High potential	54	29
Medium potential	186	13
Low potential	56	0
Rated cohort	296	42

The nine-box high-potential column converted at more than half, and the 25 high-potentials not yet promoted are the succession bench, or the next resignations, depending on what happens to them this cycle.

◆ May’s six, at the person grainthe compensation module met the same six.

May 2026 promotions	Workers
Rated 5, exceptional	2
Rated 4, exceeds	3
Rated 3, meets	1
Promotions in May	6

The six are the same six the compensation module priced against their destination bands, and the same six the change ledger logs. Ratings, money, and the event agree because they share one model.

◆ The owned answerpromotion events joined to the prior cycle.

A promotion is a staffing event; the rating is the prior cycle’s row. The join is by worker with the event dated after the cycle close, so the report always tests the rating that was actually on file when the decision happened.

-- promotions against the rating on file
SELECT r.final_rating,
  COUNT(DISTINCT r.worker_id)  AS cohort,
  COUNT(DISTINCT p.worker_id)  AS promoted_12mo
FROM fct_rating r
LEFT JOIN fct_staffing_event p
  ON p.worker_id = r.worker_id AND p.event_type = 'Promotion'
 AND p.event_date BETWEEN r.cycle_close AND DATE(r.cycle_close,'+365 day')
WHERE r.cycle_id = '2025-ANNUAL' AND r.status = 'Completed'
GROUP BY 1

Use case

Problem

Promotions are defended one by one while the pattern goes unmeasured: nobody knows if the process follows the ratings or the ratings follow the org chart.

What we build

Promotion events joined to the prior cycle’s ratings and nine-box placement, with the passed-over high-potentials counted beside the promoted.

What you get

The promise tested: 30 of 42 promotions on the top two bands, the twelve exceptions listed for explanation, and the passed-over bench named before it resigns.

Do your promotions follow your ratings?

We build the join that tests the promise every cycle.

Talk to us

Terms on this page

promotion event: A dated staffing event; the fact being tested.
prior-cycle rating: The rating on file when the decision was made.
conversion: Share of a group promoted within the window.
passed over: Rated ready, not promoted; the attrition predictor.
exception: A promotion off the expected bands; owed one sentence of why.