Favouritism in performance reviews does not usually start as a conscious decision.

It starts with a manager who knows some team members better than others. Who socialises with some employees and not with others. Who interprets ambiguous performance the way a friend would and the way a stranger would not. Over time, those small interpretive differences compound into significant rating gaps that have nothing to do with actual performance.

The problem is structural, not moral. The solution is structural too. This article covers the preventive design changes that make favouritism harder to sustain, before it damages trust and before your best people start looking elsewhere.

Design 1: Structured self-assessment before the manager rates

When employees submit a self-assessment before the manager rates them, two things happen. First, the manager is forced to respond to the employee's account of their own performance, which creates an obligation to engage with evidence the employee provides. Second, discrepancies between self-assessment and manager rating are visible to HR and trigger a conversation.

Without self-assessment, the manager's rating is an unchecked first draft. With it, the rating is a response to a shared starting point.

Design 2: Multi-rater feedback before finalisation

A manager's rating of a favoured employee is almost always more positive than the same employee's peer ratings. Multi-rater feedback surfaces that discrepancy. When the gap between manager rating and peer feedback is significant, HR has a concrete data point to investigate.

Talstack's 360 Feedback feature enables this by collecting structured peer, manager, and self-feedback against defined competencies. The output gives HR visibility into rating consistency before any formal review is finalised.

Design 3: Evidence requirement for above-expectations ratings

The most direct structural intervention against favouritism is a rule: any rating of above expectations requires a specific, written example before it is accepted into the final record. Generic praise ("she is a great team member") does not qualify.

When managers know they must justify above-expectations ratings with specifics, the most obvious favouritism is self-corrected before calibration because the manager cannot produce the evidence.

Design 4: Calibration with a fairness lens

Beyond checking for bias outliers, calibration should include an explicit fairness check: are there demographic patterns in the rating distribution that suggest systemic favouritism? If all the above-expectations ratings go to employees of the same gender, ethnic background, or tenure band, that is a fairness signal that requires investigation.

This check requires HR to have some demographic data attached to the rating distribution. In African companies where this data may be sensitive, the analysis can be done at a team level without exposing individual data.

Design 5: Manager rotation and cross-functional observation

In small organisations where an employee has been managed by the same person for three or more years, the risk of favouritism is highest. The manager has had time to build a strong positive impression that may not be fully evidence-based.

Where possible, cross-functional input should be sought for employees who have limited visibility to other leaders. This does not require formal rotation, but it does require HR to actively seek out people who can provide additional perspective on a long-tenured employee's performance.

Quick checklist: favouritism prevention

• Self-assessment required before manager ratings
• 360 or peer feedback collected before finalisation
• Written evidence required for all above-expectations ratings
• Calibration includes demographic fairness check
• Dispute process is publicly known and genuinely neutral
• Post-cycle monitoring tracks whether disputing employees receive negative consequences

Frequently asked questions

How do you address favouritism that has already damaged team trust?

Three actions in sequence: acknowledge it directly in a team context if it is widely known (vague denials make it worse), revise the process so the structural conditions that allowed it are visibly changed, and demonstrate in the next cycle that the new process produces different outcomes. Trust is rebuilt through consistent experience, not promises.

What is the difference between favouritism and a manager knowing their team well?

A manager who rates a well-known team member highly because they have more evidence of that person's performance is not showing favouritism, they are benefiting from proximity to evidence. Favouritism is when the positive evidence is interpreted more charitably, gaps are overlooked more generously, or the rating standard applied is more lenient for some employees than others. The difference is in the standard applied, not the evidence available.