Talent reviews
What is the 9-box grid? How to use it for talent reviews
The grid is only ever as good as the conversation around it. What the two axes measure, what each of the nine boxes means, and how a placement becomes a succession plan.
By the CLEAR Talent team14 min read
Key takeaways
- The grid exists to separate two questions a single rating blurs: how someone performs now, and how much headroom they have for a different role.
- Performance is evidenced from work you have already done. Potential is a judgement — define it in observable terms, in writing, before anyone rates it.
- A box is a prompt for a conversation, not a verdict. Nothing useful happens until every placement has an action, an owner and a date.
- A well-populated top-right box is not a succession plan: “ready for more” is not the same as “ready for this role”.
- Calibrate placements across managers and revisit them on a cycle, or the grid preserves last year’s opinions indefinitely.
The 9-box grid has outlived several confident predictions of its demise, and the reason is straightforward: it asks two different questions about a group of people, on one page, at the same time. How well is this person doing the job they hold — and how much headroom do they have for a different one?
It also fails in predictable ways. Potential gets rated as though it were a second performance score, boxes harden into labels people carry for years, and the grid that took an afternoon to fill in is never opened again. This guide covers what each axis actually measures, what all nine boxes mean, a template you can copy, and how to get from a placement to a succession plan somebody owns.
What the 9-box grid is, and what it is for
A 9-box grid is a three-by-three matrix. One axis carries current performance, usually rated low, moderate or high. The other carries potential — the judgement about whether someone could take on a larger or different role, and how soon. Everyone discussed lands in one of the nine boxes, and the grid shows the whole group at once. You will also see it called the nine-box talent matrix or the performance-potential matrix; the mechanics are the same.
The value is in the separation. A single performance rating collapses two things that behave quite differently: the specialist who is excellent in a role they have no wish to leave, and the person who is merely solid today because they are six months into a job that stretches them. Rank both on one scale and they look similar. Place them on two axes and they sit in opposite corners, which is a far more useful starting point for deciding what each of them needs next.
The grid is a discussion tool, and it is at its most dangerous when treated as though it produced decisions. A placement is a summary of what a room believed on a particular afternoon. That is a reasonable place to begin a conversation and a poor place to end one.
A box is where a conversation starts. It is not a verdict, and it is not a permanent property of a person.
The two axes: one is evidenced, the other is judged
Treating both axes as the same kind of rating is the most common structural mistake, and everything downstream inherits it. Performance is retrospective and evidenced: goal results, review ratings, competency assessments and feedback from the period. Most organisations already hold all of it, and the grid should read that evidence rather than reinvent it on the day.
Potential is a prediction, and predictions about people are where bias does most of its work. Left undefined, potential quietly defaults to visibility — proximity to decision-makers, confidence in a meeting, resemblance to the people doing the rating. The answer is not to drop the axis. It is to say, in writing and before the session, what this organisation means by it.
- Ability to learn
- Does this person pick up unfamiliar work quickly, and do they change how they operate when the evidence says they should? The useful examples are the ones where somebody was wrong and adjusted, not only the ones where they were right.
- Aspiration
- Does the person want a larger or different role? Willingness to relocate, to lead or to step up belongs on the grid as a stated fact rather than an assumption — and it is a question to ask the person, not about them.
- Behaviours that would scale
- How someone works under pressure, how they treat people who cannot help them, whether colleagues seek them out. This is where competency assessments and 360° input earn their place, because the manager is not the only person who has seen it.
- Evidence of stretch
- Has the person done something meaningfully harder than their job description, and how did it go? A potential rating with no stretch assignment behind it is a judgement about someone nobody has yet seen tested.
Whatever definition you settle on, put it in front of the room during the session. A shared definition is what turns nine boxes into a common language rather than nine labels each manager reads slightly differently.
It is worth being explicit about one more thing: potential describes suitability for a different role, not the worth of a person. Someone placed in the low-potential column is not failing. They may be doing exactly the work the organisation needs from them, at a standard that would be expensive to replace.
What each of the nine boxes means
The labels below are the conventional ones — use whatever language your organisation already understands. What matters is that every box carries an agreed action, because a grid where six of the nine boxes have no response attached is a picture rather than a plan.
Template
9-box grid template: what each box means and what follows it
Rows are grouped by performance band; the first column is the potential rating. Each line pairs a placement with the action it should trigger.
| Box | Potential | What the placement says | Action that follows |
|---|---|---|---|
| High performance | |||
| Star | High | Delivers now and has headroom for a bigger role | Stretch assignment, succession candidate, retention plan |
| High performer | Moderate | Strong in role with room to grow within or near it | Broaden scope, develop towards a named next role |
| Trusted professional | Low | Excellent where they are, no appetite or headroom to move | Recognise, retain, use as mentor and subject expert |
| Moderate performance | |||
| Emerging talent | High | New or stretched, showing clear signs of headroom | Coach closely, set a defined stretch task, review in six months |
| Core player | Moderate | Dependable — most of the organisation lives here | Keep developing in role; do not treat the middle as a problem |
| Solid performer | Low | Meets the standard, unlikely to move beyond it | Maintain performance, keep skills current, reassess if the role changes |
| Low performance | |||
| Enigma | High | Signs of headroom, results not there yet | Diagnose the cause: role fit, clarity, support or motivation |
| Inconsistent performer | Moderate | Falling short of a standard they could meet | Name the gap specifically, agree support and a date to review |
| Underperformer | Low | Not meeting the standard, no current evidence of headroom | Address performance directly through your normal process |
The labels are conventional rather than prescriptive — many teams use numbers or their own wording instead. Select the table to paste it into a spreadsheet, or download it below.
Two boxes cause most of the argument. Low performance with high potential is a diagnosis waiting to happen: the same placement fits someone in the wrong role, someone promoted without support and someone who has simply stopped trying, and each needs a different response. High performance with low potential is the box organisations handle worst — it usually holds the people who would be hardest to replace, and treating “low potential” as a polite way of saying “no longer of interest” is how they end up taking a recruiter’s call.
Resist the pull of the top-right corner more generally. Most organisations run on the middle of the grid, and a talent process that spends all of its attention on one box in nine is not really a talent process.
Running the talent review around the grid
The grid is the artefact; the session is the work. Most of what makes a placement defensible happens before anyone is put in a box.
Agree the population and both definitions
Decide which group is being discussed and confirm what each axis means in writing. If the definition of potential is still being debated during the meeting, placements made before the debate are not comparable with the ones made after it.
Bring performance in — do not re-rate it
Performance should arrive from the completed review cycle, ideally after calibration. Re-rating it in a talent review quietly undoes the calibration you already ran and introduces a second, undocumented standard.
Place people individually, then compare
Ask each manager for a first placement and their reasoning before the room hears anyone else’s. Placements offered sequentially converge on whoever spoke first and most confidently.
Work the disagreements, not the whole grid
Spend the time on the people whose placements differ between raters, who have moved a box since the last review, or who sit on a boundary. Everyone else can be confirmed quickly.
Record the reason next to the box
A placement with no rationale attached is indistinguishable from a guess six months later — and the rationale is the first thing anyone asks for when a promotion decision is questioned.
Leave with actions, not a picture
Everyone discussed should leave the session with a next step, a named owner and a review date. The exercise is finished when the actions are, not when the boxes are full.
A talent review has the same failure modes as a calibration session, for the same reasons: the most confident advocate, the most recent memory, the person discussed last. If you already calibrate performance ratings, run the talent review with the same discipline and much of this comes for free.
Common pitfalls, and how to avoid them
- Potential rated as a second performance score
- If the two axes correlate almost perfectly, the room is rating the same thing twice and the grid has collapsed into a ranked list. A neat diagonal of placements is a warning sign, not a result.
- The box becomes a label
- A placement describes a moment. Once “she’s a box nine” enters the vocabulary it follows people into decisions they were never placed for, and the annual refresh becomes a formality that confirms last year’s language.
- Distribution used as a quota
- Spread is a useful diagnostic: if four in five people land in the top row, the performance standard is the problem. Imposed as a quota, it turns placement into an allocation exercise and costs you the managers’ trust in the whole process.
- No written definition of potential
- Without one, potential defaults to visibility and nobody in the room can explain why two similar people were placed differently. That is also the version of the grid least likely to survive a challenge.
- A one-off exercise
- A grid built once and never revisited preserves a set of opinions well past their expiry date. Refresh it on a cycle and look at movement between refreshes — the direction of travel usually tells you more than the box does.
- A process nobody can describe
- Organisations differ on whether individuals are told their placement, and both positions are defensible. What is hard to defend is secrecy about the process itself: people should know that talent reviews happen, what the axes mean and how the output is used, even where the grid stays confidential.
From a box to a succession plan
The grid answers a question about a group. Succession planning answers a question about a role, and conflating the two is where most talent reviews lose their value. A well-populated top-right box is not a succession plan, because “ready for more” is not the same as “ready for this”.
Getting from one to the other means changing the unit of analysis. Start from the critical roles rather than from the people: for each one, who are the candidates, how ready is each of them against what that role actually requires, and what would close the distance? Readiness bands — ready now, ready in one to two years, ready in three or more, not a successor for this role — do more work here than the grid, because they are anchored to a specific job rather than to general headroom.
Then compare the candidates for a role like with like: readiness against that target role, performance trajectory rather than last cycle alone, multi-rater input on the behaviours the role needs, and whether the person actually wants it. Whatever the evidence set is, it must be the same for everyone in the pool. A comparison where one candidate arrives with competency data and another with a manager’s narrative measures preparation rather than readiness.
Finally, treat a role with a single name against it as a finding rather than a plan. Single-successor roles, and critical roles with no credible candidate at all, are the output of the exercise that most deserves a leadership team’s attention.
What actually changes: development actions
Everything above is preparation for the only part that changes an outcome — what people do differently afterwards. For most of the grid that means development actions specific enough to work on: attached to a named capability, with an owner, a date and an agreed idea of what progress would look like.
The most useful actions are usually work rather than training. A stretch assignment, a project with real consequences, deputising while someone is on leave, a piece of the manager’s job handed over for a quarter. These also generate the evidence the next talent review would otherwise be missing, because a potential judgement made after someone has been tested is a different kind of judgement from one made before.
Where the gap is genuinely a capability gap, size it against the role being planned for rather than the role being held, prioritise the two or three that matter most, and reassess on the same scale next cycle. That is what makes it possible to say whether any of this worked.
The 9-box grid in CLEAR Talent
CLEAR Talent’s succession planning includes an interactive 9-box grid with configurable axes, live filtering by role family, region or diversity dimension, and drag-and-drop placement where every move is written to an audit log. It draws on the competency assessments, 360° feedback and performance scores the review cycle already produces, rather than a separate dataset assembled for the meeting.
From there the grid connects to the rest of the plan. Readiness ratings — ready now, ready in one to two years, ready in three or more, or not a successor — each carry the specific competency gaps that would need to close to move up a category, and an assessment drafts an individual development plan from the competencies that matter most, ranked by the size of the gap. Candidates for the same role can be compared side by side against the same evidence. Dashboards flag critical roles with fewer than two successors, single-point-of-failure positions and diversity gaps in the pipeline, and the pack exports to PDF or Excel for a board or nomination committee.
None of that decides who gets the job. What it removes is the part of the decision that was being made from memory, from a deck assembled the week before, and from whoever happened to be discussed last.