9-Box Matrix Without Labels: How to Separate Performance From Potential and Calibrate Criteria Before Deciding

The 9-box matrix works best when an organization treats it as a conversation framework, not a labeling system. The design is simple: one axis represents performance in the current role and the other represents potential to take on broader or more complex responsibilities in the future. Problems begin when the two axes are confused, when each manager uses a different definition of “potential,” or when a box becomes a verdict about the person.

A safer way to use the matrix is to reverse the usual order: define criteria, evidence, and boundaries first; calibrate examples next; only then place names. This approach is consistent with talent-review materials that recommend separating performance from potential, using objective data and observations, and revisiting outcomes over time rather than turning the 9-box into a permanent label.

What the 9-box actually compares

Performance answers a retrospective question: how has the person delivered what was expected in the current role, considering results, quality, relevant behaviors, context, and enough time for observation? Lloyd’s talent-review toolkit treats performance as a measure tied to the current role and notes that time in role and experience should be considered.

Potential answers a different question: what evidence suggests that the person can take on, within a time horizon defined by the organization, broader, more complex, or different responsibilities? Lloyd’s lists factors such as motivation to lead, formative experiences, leadership competency, logic and reasoning, learning agility, and self-awareness. In another formulation, the Center for Creative Leadership associates high-potential talent with the ability, organizational commitment, and motivation to rise to and succeed in more senior positions. These definitions are not universal, but they illustrate a central point: potential needs its own criteria; it should not be inferred automatically from current performance.

Define the criteria before placing names

A three-level scale is comparable across functions only if each level has minimum anchors. “Low, medium, and high” are not enough. Before the meeting, write down what each level means and which observation period will be used. For performance, the reference can be the current cycle of goals and responsibilities. For potential, the organization needs to define potential for what: people leadership, greater technical complexity, cross-functional scope, commercial responsibility, or another type of expansion.

It is also worth defining a time horizon. Some organizations assess readiness for the next move; others assess capacity to grow two or three levels over several years. Mixing horizons creates inconsistency. If one manager interprets “high potential” as “ready for promotion now” and another interprets it as “could go far in the future,” the matrix stops comparing people and starts comparing different standards.

Do not force someone into a box when evidence is missing

Being new in role, a recent change in scope, limited exposure to complex projects, or the absence of appropriate observers can make classification premature. In such cases, “insufficient evidence” is more useful than filling the grid by obligation. Lloyd’s explicitly notes that someone may be too new in a role for performance to be assessed safely.

Build an evidence pack before the meeting

The best antidote to a reputation-based discussion is to require each assessment to come with evidence. A concise pack can include:

• period goals and deliverables, including quality and impact;
• feedback from internal customers, peers, and stakeholders when relevant;
• examples of observable behaviors rather than generic adjectives;
• experiences in new, ambiguous, or more complex situations;
• meaningful changes in context, resources, or scope;
• time in role and enough time to observe a performance pattern;
• the person’s aspiration and interest in broader responsibility;
• previous development actions and what changed afterward.

Avoid turning the meeting into a contest of narratives. Statements such as “very strategic,” “has executive presence,” or “everyone likes her” need to be converted into verifiable examples. Lloyd’s itself distinguishes evidence of potential from weak criteria such as fit, likeability, second-hand perception, and unsupported assertions.

Calibrate with questions, not advocacy for names

Calibration does not mean forcing every function to produce the same distribution of people. It means making criteria comparable and challenging differences in rating standards. SHRM notes that calibration meetings can reduce inconsistency but can also introduce new bias when poorly structured; it recommends clear criteria, rubrics, preparation, time discipline, balanced participation, and documentation of decisions.

A practical meeting sequence is:

1. restate the definitions of performance, potential, time horizon, and scale;
2. use two or three pre-discussed anchor cases to establish the standard;
3. ask the manager for the evidence supporting each axis separately;
4. invite only people with sufficient exposure to the employee to add or challenge evidence;
5. record why a placement was maintained or changed;
6. revisit cases that sit on the boundary between levels;
7. review the final pattern for signals that deserve another check.

The facilitator should stop vague personality comparisons and bring the discussion back to criteria. If someone says, “she doesn’t look ready,” the next question is, “ready for what scope, based on what evidence?” If someone says, “he is clearly high potential,” ask, “which behaviors and experiences support that prediction?”

Use bias as a set of checkpoints

Potential often depends more heavily on judgment than performance, so it deserves extra safeguards. Useful risks to check include recency, halo effects, similarity to the evaluator or current leaders, unequal visibility, preference for particular communication styles, and the effect of unequally distributed opportunities.

Aggregate analysis can also surface important questions. If certain groups appear systematically less often at the highest levels, that does not prove bias by itself; it is a signal to re-examine criteria, access to visibility, and consistency of evidence. Lloyd’s recommends using diversity data to challenge the proportionality of outcomes and reviewing whether decisions were supported by objective evidence.

Boundary cases that test the quality of the criteria

High performance and moderate or low potential

This should not be treated as failure. Someone can create enormous value as a specialist and may not want, or may not yet show readiness for, broader scope. The mistake is assuming that every outstanding performer should end up in people management. The action may be deeper technical development, recognition, broader influence without a hierarchical move, or an expert career path.

Performance is still moderate and potential is high

This can happen, for example, when someone is new in role and has not yet had time to turn capability into consistent results. But “potential” should not be used to indefinitely excuse performance issues. The plan should separate learning curve, contextual constraints, and a genuine execution gap.

High capacity but little aspiration for promotion

Potential is not automatic permission to move someone. Interest and motivation matter. Before designing a succession move, confirm what the person wants to develop and which responsibilities they actually want to take on.

Limited exposure and strong opinions

When participants know little about the person’s work, the right answer may be “I do not have enough evidence.” Repeating third-party perceptions adds noise, not precision.

After the meeting: turn placement into action

The value of the 9-box is not the box; it is the development decision that follows. OpenStax notes that after a talent review, leadership should discuss development actions and that outcomes feed succession planning and the identification of talent gaps. Lloyd’s likewise recommends that the tool lead to development plans, moves, projects, mentoring, coaching, and other opportunities that match the observed need.

Instead of names such as “star,” “solid player,” or “low potential,” record questions and actions. Examples: “what experience is still missing to test potential?”, “which result needs to become consistent?”, “is an expert path better than a management promotion?”, “which higher-complexity project could generate new evidence?”. This reduces the risk that a box turns into an identity.

The placement should also have an expiration date. Lloyd’s describes the 9-box as a snapshot in time and recommends reassessment over time. A role change, new manager, critical experience, sustained improvement, or strategic shift may justify a new assessment before the next annual cycle.

Checklist before deciding

• Do performance and potential have different written definitions?
• Is the potential time horizon clear?
• Do the three levels of each axis have observable anchors?
• Is there evidence from the full cycle, not just recent events?
• Has the manager separated facts, interpretations, and hypotheses?
• Has the person’s aspiration been considered where relevant?
• Can low-exposure cases remain as “insufficient evidence”?
• Does the meeting record the rationale for classification changes?
• Does the group review aggregate patterns without turning proportionality into a quota?
• Does every placement end with a development, support, or experience action?

In summary

A useful 9-box starts before the grid. Common definitions, evidence, and boundary examples come first; calibration comes next; names come last. The matrix can help structure conversations about performance, potential, succession, and development, but it does not turn judgment into an objective prediction. The less the organization relies on labels and the more it documents criteria, evidence, uncertainty, and actions, the more likely the tool is to support better decisions rather than merely create an appearance of precision.

Sources for further reading

• Lloyd’s — Talent review and succession planning toolkit.
• Center for Creative Leadership — High-potential talent.
• SHRM — How calibration meetings can add bias to performance reviews.
• AIHR — 9 Box Grid: A Practitioner’s Guide (2026).
• OpenStax — Talent Development and Succession Planning.