A **9 box talent assessment** is a talent-review method that places employees on a grid using current performance and assessed future potential. Each axis has three levels, creating nine possible positions. The grid can help you discuss development and succession, but a placement reflects a judgment to examine, not an objective measurement of someone's future.

The useful output is a decision: what opportunity, support, or further evidence does this person need? A neatly populated matrix without those decisions adds little. This guide gives you a neutral grid, an evidence worksheet, and a process for challenging ratings before they shape careers.

## What does a nine-box talent grid measure?

A common nine-box grid compares performance in a current role with potential for a defined future role or scope. It separates two questions that managers often blur: “How well are you doing this job?” and “What evidence suggests you could succeed at different work?”

[Oracle's succession-planning documentation](https://docs.oracle.com/cd/E18727_01/doc.121/e16346/T536783T603430.htm) describes a performance-versus-potential matrix with three levels on each axis. It also documents other matrix types. There is no single universal set of cell names, numbering, or career decisions you must adopt.

For a small company, start with the decision you need to make. You might be planning technical leadership, identifying a successor for an operations role, or deciding which skills to develop before recruiting. Those purposes need different evidence.

Do not use a general question such as “Who has leadership presence?” as your potential scale. A future engineering manager needs evidence about delegation, feedback, and coordinating people. A future staff engineer may need evidence about technical direction and influence across teams. Neither path is inherently more valuable.

The grid is also different from an external candidate scorecard. Existing employees have work histories within your organization; applicants bring different evidence and have not had the same opportunities. Use [structured hiring scorecards](/blog/skills-based-hiring-structured-scorecards) for the role you are filling, rather than giving applicants speculative nine-box labels.

## How should you define performance and potential?

Define performance against agreed expectations and potential against specific future responsibilities. Write the definitions before managers rate anyone, then record evidence and uncertainty separately.

### Performance: contribution in the current role

Ask what the employee was expected to achieve, what happened, and under what conditions. Include the quality of the work and relevant behaviors, not just output volume.

A missed launch has a different meaning when the scope changed repeatedly, a dependency failed, or the person inherited an undocumented system. That context does not erase accountability. It stops you from mistaking unlike situations for comparable results.

Possible performance descriptions are:

- **Below agreed expectations:** material expectations remain unmet, with examples and context recorded.
- **Meets agreed expectations:** the employee delivers the outcomes and behaviors their role requires.
- **Exceeds agreed expectations:** the employee makes a sustained contribution beyond the role's agreed scope, supported by specific examples.

These are starting definitions to adapt, not a validated rating instrument. Write the evidence so someone outside the employee's immediate team can understand it.

### Potential: evidence for a particular next scope

Define what the potential is for. “Potential to lead a migration involving several teams” is assessable in a way that “high potential” is not.

Separate ability, opportunity, and aspiration. Someone may have the relevant skills but not want people management. Another person may be interested but have had no chance to demonstrate the work. “Not yet observed” should not quietly become “cannot do it.”

[A 2018 working paper by Benson, Li, and Shue on promotions](https://www.nber.org/papers/w24343) examined sales workers at 214 firms and found evidence consistent with employers prioritizing current job performance when promoting people, at the expense of other observable characteristics associated with managerial performance. That sales context does not diagnose your team. It does give you a reason to avoid treating strong current results as sufficient evidence for a different job.

## What can you put in each of the nine boxes?

Use neutral descriptions tied to a possible development action. The table below is an **example review worksheet**, not a standard you must follow or a ranking of human worth.

Here, “future scope” means a responsibility you have defined in advance. This is a **readiness-focused adaptation** of the grid: it asks what the current evidence supports for that scope, rather than claiming to measure a person's long-term capacity to grow. Record career aspirations separately; wanting a different path is not a low rating.

| Assessed readiness for future scope | Below current expectations | Meets current expectations | Exceeds current expectations |
| --- | --- | --- | --- |
| Strong evidence of readiness | Understand the current-role gap before adding scope | Consider a supported stretch assignment | Discuss a credible next opportunity |
| Some evidence; development needed | Support current delivery and test a specific hypothesis | Agree one development objective | Test the next scope without assuming readiness |
| Evidence currently supports present scope | Clarify current expectations and needed support | Develop an agreed skill within the current role | Recognize expertise and explore a chosen development path |

Keep a **not assessed** option outside the matrix. Use it when the person is new, expectations were unclear, or there is too little relevant evidence. Forcing a placement creates apparent certainty without improving the decision.

Avoid attaching automatic actions to cells. A lower performance rating does not by itself establish a dismissal case. A high/high placement does not promise promotion, a pay increase, or an available role. Those decisions need their own evidence and process.

Explain those boundaries to managers and employees. Otherwise a development exercise can become an unofficial reward system whose rules nobody has agreed.

## How do you run a useful nine-box calibration?

Collect evidence independently, examine disagreements together, and leave with specific actions. Calibration is a way to examine judgment; agreement in the room does not prove the judgment is correct.

The [Australian Public Service Commission's talent segmentation process](https://www.apsc.gov.au/initiatives-and-programs/aps-talent-management/toolkit-segmentation) includes career conversations, moderation of placements, development actions, and continuing review. Its public-service framework is an example of a complete process, rather than proof that a particular matrix improves outcomes everywhere.

### 1. Define the decision and comparison group

Write down why you are running the review. Name the future scope under consideration, the period covered, and which roles can reasonably be compared.

Do not compare a new manager rebuilding a team with a long-established manager using only delivery totals. Explain the differences in scope, resources, and time in role. If the comparison needs too many caveats, discuss the roles separately.

### 2. Ask about career aspirations

Speak with the employee before the meeting. Ask what work they want more of, what they want to learn, and which future responsibilities interest them.

Do not infer aspiration from willingness to relocate, work late, or speak frequently in meetings. State the actual role requirement and discuss it directly. A preference for a specialist path is a career choice, not a deficit.

### 3. Write an evidence note before assigning a cell

Use this compact worksheet for each person:

| Field | What to record |
| --- | --- |
| Current role and review period | The expectations against which performance is assessed |
| Performance evidence | Specific outcomes, work quality, and relevant behaviors |
| Context | Resources, changes, dependencies, and opportunities available |
| Proposed future scope | The actual responsibility being considered |
| Readiness evidence | Observed work relevant to that future responsibility |
| Missing evidence | What has not been tested or observed |
| Employee aspiration | Their stated interests, with the conversation date |
| Proposed action | An opportunity, support, or further observation |
| Owner and review date | Who will make the action happen and revisit it |

A note such as “great attitude, leadership material” is not enough. Replace it with an observed behavior and explain why it matters to the proposed work.

### 4. Challenge the proposed placement

Have managers prepare their notes before hearing the most senior person's view. In the meeting, ask what supports the rating, what contradicts it, and what would change it.

Compare the evidence standard across people. If one employee needs to have performed the future role already while another receives credit for confidence or promise, the standard differs. Record that disagreement instead of resolving it through louder advocacy.

You do not need a predetermined number of people in each box. A distribution target can push the meeting toward fitting people to a pattern rather than assessing the evidence you have.

### 5. Agree an action and follow through

Choose an action that matches the uncertainty. That might mean a supported project lead assignment, clearer current-role expectations, mentoring, or time to develop a technical specialty.

Describe the support as clearly as the expectation. “Demonstrate leadership” is vague. “Lead the migration planning meeting, with the engineering manager coaching preparation and reviewing the decision record afterward” describes an opportunity someone can actually receive.

Keep the employee conversation separate from the managers' private calibration notes. Explain relevant feedback, the agreed development action, and how progress will be reviewed. Follow your organization's rules for access, retention, and correction of employee records.

## What does a worked nine-box example look like?

A useful example shows how evidence changes the action, rather than how quickly you can assign a label. The following scenario is fictional and illustrates the worksheet.

Maya is a backend engineer who reliably delivers work in her team's current scope. Her manager proposes “high performance, high potential” because Maya recently resolved a difficult production problem.

The proposed next role is engineering manager. During calibration, another manager asks what the incident shows about coaching, delegation, and supporting other people's work. The answer is: very little. It is strong technical evidence, but it does not settle readiness for people management.

The career conversation adds another fact: Maya wants to deepen her technical influence and has not expressed an interest in managing people. The team revises the action. Maya will lead a design review with a senior engineer's support and explore the technical career path.

That correction makes the review useful. It recognizes contribution, respects aspiration, and avoids treating management as the default reward for expertise.

Now consider an employee whose performance is below expectations but who appears capable of wider work. Before assigning a stretch opportunity, ask whether the current expectations were clear and whether the role matches the person's strengths. Do not use a positive potential label to avoid a difficult performance conversation, or a performance problem to erase relevant strengths.

## Where can nine-box talent assessment go wrong?

The biggest risk is converting a subjective assessment into a durable fact. Look closely at unequal opportunities, vague language, and labels that follow people after the evidence changes.

In a [2026 American Economic Review study](https://www.aeaweb.org/articles?id=10.1257/aer.20220831), Benson, Li, and Shue analyzed 29,809 management-track employees at one large retail chain. Women received lower potential ratings despite higher performance ratings; differences in potential ratings accounted for approximately half of the gender promotion gap in that setting. This is evidence from one employer, not an estimate for every organization. It is a serious reason to challenge how potential is judged.

Review comments for words such as “polish,” “presence,” and “fit” that lack a concrete role requirement. Ask whether people received comparable opportunities to demonstrate the behavior. Track who gets development assignments as well as who receives favorable ratings.

Also watch for these mistakes:

- **Recency:** one visible success or failure outweighs the agreed review period.
- **Visibility:** time in meetings is mistaken for contribution, particularly across remote and office-based teams.
- **Halo effects:** strength in one area becomes assumed strength in unrelated work.
- **Manager advocacy:** a confident presenter secures more opportunity for their team.
- **Permanent labels:** an old placement survives after responsibilities and evidence change.

These are prompts for scrutiny, not a claim that a checklist removes bias. If you manage a distributed team, make the evidence-gathering process explicit across [locations and time zones](/blog/hiring-distributed-remote-team-across-time-zones).

## Common questions about nine-box talent reviews

Use the grid only when it clarifies a real decision. You can run a useful development conversation without placing everyone in a matrix.

### Is nine-box an employee performance review?

It can use performance-review evidence, but it adds a separate judgment about future scope. Keep current expectations, development, promotion, and compensation decisions distinguishable. Employees should understand which conversation is taking place.

### Should everyone aim for the high-potential box?

No. The answer depends on the future role, the employee's interests, and the evidence available. Strong specialists and dependable current-role contributors need recognition and development too. [CIPD's talent management guidance](https://www.cipd.org/uk/knowledge/factsheets/talent-factsheet/) describes both targeted and inclusive approaches to developing talent.

### How often should you update the grid?

Choose a cadence that matches your decisions, and revisit an assessment when scope or evidence materially changes. There is no universal interval established by the sources in this guide. Development actions should have their own review dates so they do not wait for the next company-wide exercise.

### Does a high rating guarantee a promotion?

No. Readiness, employee interest, role availability, and the organization's promotion process are separate questions. [OPM's succession-planning approach](https://www.opm.gov/agency-services/talent-management-services/strategic-consulting/workforce-and-succession-planning/succession-planning/) focuses on key positions and future readiness. A succession discussion is preparation for a possible need, not an offer.

## Turn a workforce gap into a clear hiring process with Kit

A good talent review can show that you need development, a different allocation of work, or an external hire. Make that choice before opening a role. Do not turn an employee's confidential potential rating into an applicant-selection criterion.

If hiring is the right decision, define the work and the evidence you will use to assess candidates. Kit supports job postings, hiring stages, interviews, and team reviews. It is a recruiting system; it does not provide a nine-box employee performance or succession-planning module.

Use the nine-box worksheet to make a better development decision, then use role-specific evidence for the hiring decision. You can explore [Kit's hiring templates](/templates) when you are ready to build that process.