9-Box Grid

Talent reviews are hard to run well. Without structure, they become loose discussions dominated by whichever manager is loudest. With structure — even imperfect structure — they force consistent discussion of who is where, why, and what the company is doing about it. The 9-box provides that structure cheaply. The trade-off is that the structure is crude — two dimensions, nine boxes, a snapshot in time — and treating it as more authoritative than it is produces decisions that don't survive scrutiny. The value is in the conversation the grid forces, not in the cell assignment itself.

Take a 2,800-person technology company running annual talent review. Each manager rates their direct reports on performance (last 12 months delivery) and potential (ability to grow into larger roles). Ratings go into the 9-box. Calibration meetings convene same-level managers to compare ratings across teams and adjust outliers. The output: named-cell placements that feed succession plans, development investment (top-right cells get expensive learning budgets), and compensation decisions. Process takes two months, runs in the HCM's talent module. Leadership reviews the aggregate grid to understand bench strength and gaps.

The grid works when it is one of several inputs, treated with humility about its limits. It fails when it becomes the primary mechanism for who gets promoted, who gets invested in, and who gets managed out. The "potential" axis is especially problematic — most organizations have no operational definition of potential, so it becomes a proxy for who the senior managers like. This drives systemic bias (the "high potential" pool often looks demographically narrow compared to the workforce). Mature organizations use the 9-box for directional conversation and couple it with objective data (project outcomes, peer feedback, skill assessments) before making consequential decisions.

In tech and professional services, 9-box is near- universal in companies over 1,000 employees. It's often integrated into the HCM's talent module.

In financial services, where regulatory frameworks require documented talent decisions, 9-box is part of the compliance record.

In manufacturing and healthcare, 9-box is less universal — more companies rely on competency-based or succession-specific tools.

In retail and hospitality at the hourly level, 9-box is rare and generally unsuitable — the talent review discipline for hourly workers runs on different tools (readiness matrices, skills certifications).

AI changes 9-box in 2026 by adding data that was previously unavailable. Performance data from project outcomes, peer sentiment from feedback tools, skill trajectory from learning data, and engagement patterns all flow into the talent review as richer evidence. The 9-box assignment stops being a manager's opinion and starts being a manager's synthesis of data the system surfaced. Calibration meetings shift from "do we agree with Sarah's rating" to "do we agree with the evidence behind Sarah's rating." The grid itself may not survive — or may evolve to more dimensions — but the underlying discipline of structured talent review becomes much richer.

• Stack-ranking via 9-box. The grid is a snapshot and a conversation starter. Used as a forced-distribution ranking, it produces political gaming and low psychological safety.
• "Potential" as likability proxy. Without a concrete operational definition, potential becomes "who does senior leadership trust." Predictable bias follows.
• No calibration. Uncalibrated ratings mean "top-right" in Sarah's team means something different from "top-right" in Mike's team. Calibration meetings are non-negotiable.
• Treating as permanent. A person's placement at annual review is a point-in-time snapshot. Treating it as durable labels people for years beyond the relevance of the rating.
• Sharing with employees literally. "You're in the lower-left box" conversations go poorly. Share the development implications, not the cell assignment.