Performance management differs by manager when the organization has no shared standard for good performance, or has one that managers do not use, and nothing to bring ratings back in line. The fix follows an order: define the standard in observable behaviors, build it into the review process, calibrate ratings against it, and then train managers to apply it.

Why does the same quality of work earn a “meets expectations” on one team and an “exceeds” on the next? In our work with mid-sized organizations, the usual answer is that nobody gave managers a shared definition of good performance, so each manager supplied their own.

The patterns are easy to recognize once you know what to look for: people in the same or similar roles held to different standards, strong ratings that follow a manager’s liking for someone more than the work, a manager who rates everyone high, and promotions that go to people who were not ready for the next level. Each one makes later decisions harder to defend, because a rating that depends on who the manager is cannot be explained to the employee who received it or to the leaders who rely on it for promotion and succession.

Our position is that this is a system problem, and that it gets fixed in a specific order: the standard first, then the process that applies it, then calibration, and only then training for managers. A new review form, a calibration meeting, or a training program each feels like progress, yet none of them can create consistency while every manager still carries a private definition of good performance.

The table below shows where the breakdown tends to occur, what it looks like, and a question that tests for it.

Start with a standard that two managers would describe the same way

A usable standard describes good performance in observable behaviors for the role and level. A competency model with behaviors at each level is built to do this job, and the same model should guide development and succession so the organization is not maintaining a separate definition for each process. We walk through building one in how to build a leadership competency model for a mid-sized company. If a standard already exists and managers ignore it, the problem has moved to the next step.

Build the standard into the process managers already use

A standard that lives in a document and not in the review form will not be used. Put the behaviors into the review criteria, into the examples managers are asked to cite, and into the promotion criteria. For promotion and succession decisions, compare the person to the behaviors of the next level, which guards against promoting someone who excelled in the current role and is not yet ready for the next one. For development, use the behaviors of the person’s current level. Applying the same behaviors to everyone in a role also removes a very visible inconsistency, two people doing similar work under different expectations.

Calibrate against the standard

Calibration is a meeting where managers review proposed ratings together before they are final. It limits the effect of a manager who rates everyone high and a manager whose favorites rise to the top, because each rating has to be supported by examples that match the standard. Without a shared standard, the same meeting becomes a negotiation between individual definitions, and ratings tend to follow whoever makes the better case. Looking at rating distributions by manager beforehand shows where the conversation needs to start.

Then equip managers to apply it

Training comes after the standard because managers need something specific to be trained on: how to observe against the behaviors and how to give specific feedback during the year, so the review holds no surprises. Training first sends managers toward an unclear standard. This is not an argument against training; it is an argument about order. For what a program should include, see new manager training: what to include and how to know it worked.

How do you tell where the inconsistency starts?

Rating distributions by manager are the quickest signal: a manager whose ratings cluster at the top, or a team where no one’s rating ever varies, points to a standard that is not being applied. Reading the review documents adds a second signal, the quality of the evidence, since strong reviews cite specific behaviors and examples while weak ones rely on general impressions. In our experience, plenty of companies do not have either artifact in usable form. In that case the signal comes from interviews with managers about how they decide a rating, and from employee survey items on whether expectations and feedback are clear. The free OGC Talent Health Snapshot is a way to get a first read on how your talent practices are working together.

Wondering whether a rating means the same thing on every team in your organization? Let’s talk through how to find where the inconsistency starts and what to fix first.

Have a specific question about how this applies to your organization? Our FAQ page covers common questions on talent systems, leadership development, and engagement data.