9-Box Grid: A Practical Guide to Talent Reviews and Calibration

What is the 9-box grid?

A 9-box grid is a talent review tool. The horizontal axis usually shows performance: how well someone delivered in their current role over the review period. The vertical axis usually shows potential: how far and how fast they could grow into broader or more senior roles. Each axis has three bands, so every employee lands in one of nine boxes.

The format borrows from portfolio planning. McKinsey describes the original nine-box matrix as a tool that arose in the early 1970s to compare business units on industry attractiveness and competitive strength.^[1] Organizations later applied the same three-by-three logic to people, and it has since become a familiar format for talent reviews and succession discussions.

Two things make the 9-box useful. It forces a distinction between what someone has done and what they might do next, and it gives a group of managers a shared picture to discuss. Both only work if the inputs are honest, which is where most implementations struggle.

The two axes: performance and potential

Most problems with the 9-box start with the axes. If performance and potential are defined loosely, managers fill both with the same overall impression and the grid collapses into a diagonal line from bottom left to top right.

Performance: what the person delivered

Performance is the easier axis because it looks backward and can rest on evidence. Strong inputs include goal outcomes such as OKR scores, results against role expectations and the structured ratings from the year-end review. Use the same review period for everyone on the grid, and decide in advance which sources count and how they combine.

Potential: potential for what?

Potential is harder because it is a forecast. In their review of how organizations identify high potentials, Silzer and Church found little agreement within or between organizations on what the concept actually means.^[2]

The practical fix is to define potential against a destination. Potential for a larger role in the same function, potential for people leadership and potential for a senior general management role are different questions, and the same person can score differently on each. Write the definition down before the review and make every manager use it.

Keep the two axes apart

The most common error is letting performance leak into potential. A strong performer is easy to label high potential, and a newcomer who has not had time to deliver is easy to underrate. Dries and Pepermans name this confound between performance and potential as one of the typical issues that undermine assessments of leadership potential.^[3]

A simple guard is to agree on different evidence for each axis:

The nine boxes, one by one

There is no standard set of box names. Different sources call the top right box Star, Future Leader or Exceptional Talent, and some even swap the axes. What matters is that everyone in your organization uses the same labels with the same meaning. The names below are common examples, with what each box usually signals and a typical next step.

Treat each box as the start of a conversation, not a verdict. Placements change from one review to the next, and a box that follows someone for years without fresh evidence does more harm than good.

How to run a 9-box talent review

A talent review has two parts: placing people on the grid and agreeing on what the placements mean. The steps below keep both grounded in evidence.

1. Define the axes and the evidenceWrite down what performance and potential mean for this review, which sources feed each axis and how scores map to the three bands.
2. Pre-place people from dataStart from the scores rather than from a blank grid. When placements come from review data, the meeting discusses exceptions instead of rebuilding every rating from memory.
3. Calibrate as a groupManagers review the grid together, challenge placements that look out of line and move people only with a stated reason.
4. Decide an action for every boxEvery placement leads to a next step: a stretch assignment, a development plan, a retention conversation or a performance plan.
5. Follow up next cycleCompare this year's placements with last year's. Movement over time tells you more than any single grid.

Running the calibration meeting

Calibration is where the grid earns or loses its credibility, and the research on calibration committees is encouraging but not unconditional. In a study of a large multinational, committees adjusted about a quarter of initial ratings, mostly downward. They made ratings more consistent across supervisors and reduced leniency, but they also pushed ratings toward the middle of the scale.^[4]

Calibration can also change how managers rate in the first place. In a professional services firm, the calibration committee disciplined supervisors who strategically inflated their teams' ratings, and that behavior was reflected in the supervisors' own ratings and promotions.^[5]

A few practices make the meeting work:

• Send the pre-placed grid and the evidence behind it before the meeting, so the time goes to discussion rather than reading.
• Ask for evidence with every move: a result, an example or a pattern, not an impression.
• Use a facilitator who does not manage anyone on the grid and who watches for one voice dominating.
• Look at the distribution at the end. If almost everyone sits in the middle box, the group may be avoiding hard calls.
• Record each decision and its reason, so next year's review starts from what was agreed.

Try it below: drag a card to another box and watch the distribution and the decision log update while the review score stays the same.

This is how 9-box calibration in Devokr works: people start in the box their review scores point to, managers move them by drag-and-drop, and every move is recorded while the review score itself stays unchanged. In Devokr the grid is called a talent map, and you name the boxes yourself.

Where the 9-box goes wrong

The 9-box is only as fair as the judgments that feed it, and the potential axis carries the most risk.

Bias in potential ratings

The clearest evidence comes from a study of 29,809 management-track employees at a large retail chain that used a Nine Box grid. Women received substantially lower potential ratings despite receiving higher performance ratings, the difference accounted for about half of the gender promotion gap, and women went on to outperform their male colleagues.^[6]

The lesson is not to drop potential ratings but to treat them as the part of the grid that needs the most scrutiny: define them, ask for evidence and check the distribution across groups during calibration.

Boxes that turn into labels

A box assigned once can follow a person for years. People in the lower boxes may get fewer chances to prove themselves, which then makes the original placement look right. Revisit every placement each cycle and keep the reasons, so a box reflects current evidence rather than history.

Everyone in the middle

Calibration reduces leniency but can increase centrality, as the multinational study above showed. A grid where most people sit in the center box hides the differences the exercise was meant to reveal. Ask managers to justify middle placements as carefully as extreme ones.^[4]

The 9-box and succession planning

The 9-box is often the first input to succession planning, but potential is not the same as readiness. Someone can have high potential for a senior role and still need two years of development before they are ready to step into it.

Keeping readiness as a separate judgment and tracking it over time turns the grid into a working plan. That is where succession planning with OKRs helps: readiness becomes a quarterly goal with measurable progress instead of a name on a chart.

Used well, the 9-box is less a rating tool than a structured conversation. It works when the axes are defined before anyone is placed, when placements start from evidence and when the people calibrating them are willing to challenge each other.

It also works best as part of a rhythm rather than a once-a-year ritual. Regular feedback conversations between reviews give managers the evidence the grid needs and give employees a clearer picture of where they stand.