11 min read
Calibration
For: HRBPs and Total Rewards leaders running calibration sessions
How to keep leveling consistent across teams, especially in the messy boundary cases that reveal what your company actually values.
Step 1 of 5. Calibration builds a shared standard.
Step 1 of 5
Calibration builds a shared standard
Calibration is the practice of arguing toward consistency. It exists so the standard lives in the room, not in any one person's head.
The best sessions don't end with a tally. They end with a stronger shared mental model. The point of comparing roles is to make sure that next quarter's leveling questions get the same answer that this quarter's did.
Bring two roles
Two reference roles is the smallest useful unit for calibration. The comparison is what makes the level visible.
The single most useful habit I've watched work: every manager who brings a role brings two more roles for context. One clearly below the proposed , one clearly above. The conversation stops being about defending an answer and starts being about placing the role on a real ladder.
HRBP supporting a 90-person unit
P4 Advanced
High confidence.
HRBP supporting a 350-person unit, cross-region
P6 Principal
High confidence, boundary P5/P6.
| Dimension | Role A | Role B | Delta |
|---|---|---|---|
| Scope | 4 | 5 | Role B higher (+1) |
| Complexity | 4 | 5 | Role B higher (+1) |
| Autonomy | 4 | 5 | Role B higher (+1) |
| Influence | 4 | 5 | Role B higher (+1) |
| Knowledge | 4 | 5 | Role B higher (+1) |
| Impact | 3 | 5 | Role B higher (+2) |
Boundary cases reveal philosophy
Boundary cases show what your company actually values. The argument you record becomes the precedent for the next one.
Borderline roles are the ones where your company's actual values show up. A role that's defensibly P4 or P5 forces a real conversation about how much weight you give influence versus complexity. The decision matters less than the rationale you record.
I've kept the write-ups from old projects on the way managers reasoned about boundary cases. Those notes are more useful than the levels themselves, because they tell future-me how this company tends to think.
Govern exceptions
Exceptions are fine. Undocumented exceptions become precedent and erode the standard.
Exceptions happen. Someone gets hired into a role at a the architecture doesn't quite support, because the market is hot, or the team needs the hire fast, or there's a strategic moment. Hidden exceptions become precedent. Named exceptions stay manageable.
| Field | What to capture |
|---|---|
| Role | Function, family, current level, intended level |
| Reason | Hot market / strategic hire / interim / individual stretch |
| Boundary expected | Date or condition by which the gap should resolve |
| Reviewer | Who owns the next checkpoint |
Exception hygiene
- Field
- Role
- What to capture
- Function, family, current level, intended level
- Field
- Reason
- What to capture
- Hot market / strategic hire / interim / individual stretch
- Field
- Boundary expected
- What to capture
- Date or condition by which the gap should resolve
- Field
- Reviewer
- What to capture
- Who owns the next checkpoint
Worked walk-through
Walk a real boundary case end to end. Predict, compare, and document the call so the next one has precedent.
- The case: an HRBP supporting a 200-person business unit, advising the leadership team, owning talent reviews and engagement.
- Manager's ask: P5.
- Reference roles brought to the room: a P4 HRBP supporting a 90-person unit, a P5 HRBP supporting a 350-person unit with cross-region scope.
- The conversation: scope and influence sit between the two references. Knowledge and complexity look closer to P4.
- Decision: P4 with a P4/P5 boundary noted. Re-review in six months if the unit grows past 300.
- Documented rationale, not just the level.
This is also one of the calibration scenarios in the practice lab. See §10.