Organization Decide

9-Box Talent Grid

The 9-box talent grid plots people by current performance and future potential to guide development, succession, and retention decisions.

The questionGiven this person's current performance and likely potential, what is the next move, develop, retain in role, stretch, or exit?

Also called a nine-box grid, 9-box, or talent matrix. Distinct from the GE-McKinsey nine-box used on business units.

Matrix · GE and McKinsey talent practice, 1970s onward

Potential	Low performance	Medium performance	High performance
High potential	Coach or check the role	Stretch and develop	Promote and retain
Medium potential	Performance plan	Core, keep growing	Retain and deepen
Low potential	Exit or reassign	Solid contributor	Keep in role and reward

Typical next actions by cell. Performance runs across. Potential runs down from high to low, so the succession names sit in the top right.

When to use it

A succession discussion is a list of names each manager protects, and you need a shared, evidence-based sort before anyone is put on a slate.
High performers are being pushed into a leadership program they do not want, or cannot do, because potential was never separated from performance.
Retention risk is concentrated in a few people, and compensation and development are still being spread evenly.
A talent review has to produce actions, not a heat map, for development, stretch, keep-in-role, and exit.

When to avoid it

Performance ratings are not comparable across managers, and you have no time to calibrate them. The grid will sort politics, not people.
Potential has no definition. Without an agreed meaning, such as capacity to do a role two levels up within a stated time, the vertical axis is a popularity score.
The unit is small enough that the leader already knows each person and the next moves. Drawing nine boxes adds a label without adding a decision.
The organization will use the box as a brand on the person, visible in files for years. A snapshot rating treated as identity is how the tool does its damage.

Overview

The 9-box talent grid plots people by current performance and future potential so a review can choose a next move, develop, stretch, keep in role, or exit, instead of treating every high performer as a future leader. The top-right cell is a succession slate. The high-performance cells below it are often the people the business actually runs on, and they need retention, not a promotion they will fail.

The 3x3 shape was built for GE business units, not for people. The talent version has no single author and no standard cell names. It is only as good as the definitions and the calibration. If potential is a copy of performance, if everyone is a star, or if the box becomes a brand on the person, the grid is doing harm. Write actions per person, keep the grid in the room, and let people move the next time you plot.

The two axes and the cells

Performance

Delivery in the current role, against agreed objectives and the how, not only the numbers. Use the latest cycle's evidence, calibrated across managers so that one person's "high" matches another's. Performance is about this job, this year.

Potential

Likelihood of succeeding in a substantially larger role, in a defined window, given ability, aspiration, and engagement. It is not "does well today." High potential means the person could move up, not that they must. Many high performers have low or medium potential for the next job, and that is a normal, useful finding.

High potential, high performance

The succession names. Promote, give a stretch that is a real job, and retain deliberately. There should be few. A crowded top-right box is the usual sign that managers have not been willing to differentiate.

High performance, lower potential

Specialists and trusted operators who make the current results. Keep them in role, pay them well, and stop forcing them onto a leadership track they will fail or leave. Losing these people to a mis-promotion is a common own goal.

High potential, lower performance

Often new to the role, poorly managed, or in the wrong seat. Coach, check the fit, and set a time-boxed expectation. If performance does not move, potential was a hope.

Low performance, low potential

A decision, not a development plan. Exit or reassign. Leaving people here for another cycle teaches everyone else that the grid is theatre.

What it is not

The GE-McKinsey nine-box on business units. Attractiveness against strength allocates capital. Performance against potential allocates development and succession. Do not brief one as the other.
A measurement of a person's worth. The box is a current view of two ratings, both of which can move, and both of which are judgments.

The steps

Define the axes before anyone is placed

Write what high, medium, and low mean for performance in this cycle, and what potential means, including the time window and the next role. If managers cannot say those sentences the same way, stop. The output is a one-page definition, not a grid.
Gather evidence, then place

Use objectives, results, and observed behavior, not memory. Place people independently, then calibrate in a room where managers have to explain the box with facts another manager could contest.
Challenge the crowded boxes

Count the top right and the bottom left. If a third of the population is "stars," or nobody is low, the ratings are not yet differentiated. Recalibrate those cells first.
Write an action per person, not per cell

The cell suggests a default, promote, stretch, keep, coach, or exit. The person still needs a named next step, an owner, and a date. Two people in the same cell can need different moves.
Keep the grid in the review

Do not publish boxes to the wider organization, and do not leave last year's box in the file as a label. Repeat the review on a cycle, and let people move.

In practice

A commercial function of forty-two people went into its annual talent review with eleven names in the top-right box. The HR lead asked each nominating manager for the last two years of objectives, a forced ranking of "ready now for the next job," and whether the person wanted that job. After calibration, four people remained as high potential and high performance. Eight high performers were placed as low or medium potential specialists. Three were high potential and medium performance, each in a role they had held for less than a year.

The actions changed. The leadership program shrank from eleven seats to four, with named backups and stay interviews. The eight specialists were taken off the program, given a market adjustment, and told the company wanted them in role. Two people in the low-low box, both with two years of missed objectives and no disputed facts, were put on exit. Six months later the function had not lost any of the four, and had lost one specialist who had been about to enroll in a management course she did not want. That loss was the one the first grid, with eleven stars, would have missed.

What you leave with

Written definitions of performance and potential for this review.
A calibrated grid, with counts that a skeptic would believe, especially in the top right.
A next action, owner, and date for every person discussed, not only the stars.
A succession slate that is a subset of the top right, with backups, not a copy of every high performer.

Failure points

Performance copied onto potential

This year's high performers are rated high potential by default. The specialists get put on a leadership track, the true development candidates are hidden, and the vertical axis does no work.

Stars for everyone

Managers refuse to put their people anywhere but the top right. The slate is too large to act on, and nobody receives a hard message.

The box as a brand

People learn their cell, or their cell leaks. A medium-potential rating becomes an identity, development stops, and good specialists leave.

Development only for the top right

Everyone else is ignored, including the high performers the business actually runs on. Retention problems then show up in the middle of the grid, where the review spent no time.

No exit from the bottom left

Low-low stays occupied year after year. The rest of the room concludes that placement has no consequences, and the next calibration is softer still.

The failures above are usually Stakeholder Algebra: while the analysis was precise, the people who decide were not handled correctly.

Origin

The nine-box shape is older than the talent tool. In the early 1970s McKinsey built a nine-box matrix for General Electric to rank business units on industry attractiveness and competitive strength, as a way to allocate investment. That GE-McKinsey matrix is a strategy tool. It is not a people tool.

Human-resources practice later used the same 3x3 layout on people, with current performance on one axis and future potential on the other. GE's Session C talent reviews, running by the late 1970s, assessed performance, promotability, and succession, and a nine-block of people is widely associated with that process. McKinsey and other firms use a similar grid in talent reviews. Who first swapped the axes from businesses to people is not well documented, and there is no trademark on the talent 9-box. Treat any cell labels you meet as house vocabulary, not as a standard. The grid is a calibration device for a conversation among managers who have evidence. It is not a measurement of potential.