The 9-box grid: what it is, how to run it and common mistakes
By Deibson Silva · Reviewed on
The 9-box grid is a three-by-three chart that plots each person's performance against their potential, to decide where to invest in development and whom to prepare for a key role. It works when both axes are defined before you use it, when positions are calibrated as a team with evidence, and when a person makes the call, not the grid.
What is the 9-box grid?
It is a talent review tool with two axes. The horizontal axis shows performance: what the person has achieved in the period, backed by evidence. The vertical axis shows potential: the capacity to take on more responsibility or a different role in the coming years. Each axis is split into three bands, low, medium and high, and where they cross gives nine boxes.
Its value is not in the chart but in the conversation it forces. To place someone, the leadership team has to agree on what it means by good performance and by potential, and explain with facts why each person sits where they do.
It is mainly used for three things: deciding which people to invest in, preparing succession for key roles, and spotting people whose current role does not let them perform as well as they could.
What does each box mean?
Each box suggests a conversation, not a label. The table gives a common reading of all nine, with the question worth asking in each case. Box names vary from one company to another; what matters is that they describe a situation today rather than a judgement about the person.
| Performance / potential | Common reading | Next question |
|---|---|---|
| High / high | Delivers and can take on more | Which challenge or role do we prepare them for, and by when? |
| Medium / high | Capable of more than they show today | What is holding them back: the role, the support, the clarity? |
| Low / high | Capacity without results yet | Are they in the right role, with what they need? |
| High / medium | Very strong in their role | How do we widen their scope without changing role? |
| Medium / medium | Steady contribution | Which specific development helps them this year? |
| Low / medium | Below what was expected | Which goals and which support do we agree for the next period? |
| High / low | Expert in their function | How do we recognise and keep that knowledge? |
| Medium / low | Meets the needs of the role | What do they need to keep doing it well? |
| Low / low | Difficult fit today | Is there another role where they fit better, or a clear improvement plan? |
How do you rate each axis?
Performance is rated with what already exists: the review cycle, goals or OKRs, 360-degree feedback and the results of the period. The rule is that every rating rests on facts someone else could check, not on the reviewer's impression.
Potential is the hard axis, because it is about the future. Define it in writing before the session, with a few observable criteria: how the person learns when faced with something new, whether they take responsibility beyond their role, how they cope outside their area, and whether they want to grow, which you only find out by asking them. A behavioural profile can help prepare that conversation, but it never replaces evidence.
Some criteria cannot enter either axis. Age, sex, family situation, health or any other protected ground does not define anyone's potential, and using them would be discriminatory.
How do you run a calibration session?
A calibration session is the meeting where the leadership team reviews the proposed positions together and adjusts them against a single standard. Without calibration, each manager scores in their own way and the grid compares things that cannot be compared.
An order that works has five steps. First, each manager proposes positions for the people in their team, with two or three pieces of evidence per axis. Second, someone who rates nobody chairs the session. Third, the cases at the extremes and those raising most doubt are reviewed first. Fourth, every change of box is recorded with its reason. Fifth, the session ends with an agreed development action for each person, not just a box.
Repeat the calibration at least once a year and whenever something important changes in the organisation. A grid nobody reviews becomes an old snapshot that keeps making decisions.
What limits does the law set?
The 9-box grid is lawful, but its use has clear limits. In Spain, Article 24.2 of the Workers' Statute requires promotions and career progression to follow criteria and systems aimed at ensuring there is no direct or indirect discrimination between women and men. Article 17 makes employer decisions that discriminate on the grounds it lists null and void, and Article 9 of Law 15/2022 bans exclusions on those grounds in training and career progression too.
If a position on the grid is calculated automatically and a decision affecting the person depends on it, Article 22 of the GDPR applies: it protects people from decisions based solely on automated processing. And if a company uses an artificial intelligence system to decide promotions or to evaluate performance, Annex III of the EU AI Act treats it as high risk from 2 December 2027, under the timetable set by Regulation (EU) 2026/1744.
In addition, Article 64.4.d of the Workers' Statute gives the works council the right to know the parameters, rules and instructions behind algorithms that affect working conditions. A grid that relies on an automatic calculation should be explainable.
Do you tell each person which box they are in?
That depends on the company's culture, but every person should know what is expected of them and what development plan they have. Sharing the box as it stands tends to turn it into a label that weighs on someone for years; sharing what lies behind it, the facts and the plan, helps them move forward.
Either way, the grid contains personal data. Staff should be told that it exists, what it is used for and who sees it, and access should be limited to those who need it. Each person has the right to access the data held about them.
What are the most common mistakes?
The most common is mixing up the axes: rating someone high on potential because they do their current job well, when today's performance does not always predict how someone will handle a different job. The second is the halo effect, which carries a good impression from one axis over to the other.
Other frequent mistakes: using the grid to set variable pay, which mixes two different conversations; placing people without evidence; not recording the changes made in calibration; running it once and never revisiting it; and, above all, treating a box as a fixed destination, when it describes a moment.
Sources
- Spanish Workers' Statute (Royal Legislative Decree 2/2015), Arts. 4.2, 17, 24 and 64.4.d (BOE, in Spanish)
- Law 15/2022 on equal treatment and non-discrimination, Art. 9 (BOE, in Spanish)
- Regulation (EU) 2016/679 (GDPR), Article 22
- Regulation (EU) 2024/1689 (AI Act), Annex III, point 4
- Regulation (EU) 2026/1744 (digital omnibus on AI), Article 113 dates
Frequently asked questions
- What is the difference between performance and potential?
- Performance measures what was achieved in the period, with evidence. Potential estimates the capacity to take on more responsibility or a different role in future. Someone can perform very well without wanting another role, and the other way round.
- How often should you run a 9-box grid?
- At least once a year, after the review cycle, and whenever there is a significant change in the organisation or in a key role.
- Who should attend the calibration session?
- The managers who rate the people under review, someone from people or HR to chair it and, when key roles are discussed, senior leadership.
- Can a tool place people on the grid?
- It can propose a position from data, but the decision should be taken by a person with the authority to change it, as Article 22 of the GDPR requires when a decision significantly affects someone.
- Is the 9-box useful for a small business?
- Yes, in a simple form: the roles that matter most, written criteria for each axis and a short calibration session. With few people the risk of labelling is higher, so the focus should be on the development plan.

