Inclusive performance management

Inclusive performance management means designing how objectives are set, how evidence is kept and how ratings are made, so that people are judged on their work rather than on how closely they resemble the person rating them. Most organisations approach inclusion in performance management through training about bias. That is worth doing, but it treats the problem as a matter of individual awareness when much of it is structural. Bias enters performance ratings at specific, identifiable points in the process, and several of those points can be redesigned so that an individual manager's good intentions are not the only safeguard.

Where it actually enters

The first point is the criteria themselves. When a rating depends on words like initiative, presence, commitment or cultural fit, the manager has been handed a container to fill with whatever they already believe. Those words are not neutral. Visible confidence reads as competence in some cultures and as showing off in others. Long hours read as commitment unless you have caring responsibilities, in which case they read as availability. Criteria that describe outputs and observable behaviour narrow the gap considerably, not because managers stop having biases but because the rating has less room to carry them. Stanford researchers Shelley Correll and Caroline Simard found that women's reviews contained more vague feedback and fewer comments tied to specific business outcomes than men's (Harvard Business Review, 2016).

The second is recall. An appraisal written from memory is an appraisal of what the manager noticed, and people notice more of the work done by colleagues they talk to more. In a team split between people in the office and people at home, or across time zones, the difference in what gets noticed is large and systematic. Take two analysts who deliver the same results, one in the office near the manager and one working remotely through agreed project channels. At review time the first is described as engaged and proactive, the second as quiet. The review has measured visibility rather than contribution. Contemporaneous notes are a tedious discipline that fixes most of it.

The third is comparison. Managers rate people against an unstated reference, usually the strongest performer they have managed in that role, or themselves at that stage. Where that reference resembles one group more than another, the whole distribution tilts. This is the one that calibration between managers is actually for, and it only works if the calibration meeting examines the reasoning behind ratings rather than the spread of them.

What standardised criteria will not fix

Standardising the form does not standardise the judgement. Two managers applying the same scale to the same evidence routinely disagree, and the disagreement is where bias survives. The practical check is to look at outcomes rather than intentions: compare ratings, promotion rates and development opportunities by group, by manager and by location, every cycle. If the pattern is consistent in one direction, the process is producing it, whatever the criteria say. Organisations that only audit at launch find this out years later.

It is also worth being careful with the opposite error. Adjusting a rating upward to correct an aggregate pattern is unfair to the individual and corrodes the credibility of every rating in the system. The repair belongs upstream, in the criteria, the evidence and the calibration, not in the final number.

Feedback does not read the same to everyone

Direct criticism is normal and expected in some working cultures and humiliating in others. The same sentence can land as useful candour or as a public reprimand depending on where the person learned what feedback sounds like. This is well documented, and the usual advice to simply be direct is not universally good advice. Harvard Business Review's work on feedback across cultures is a reasonable starting point for anyone managing a distributed team.

Multiple feedback sources help here for a reason that is often missed. The point is not to average opinions. It is that a single source means a single set of blind spots, and a colleague who works with someone daily sees work a manager never observes. That is most valuable where the manager and the employee have least in common.

What inclusive performance reviews look like

In an inclusive performance review, every rating can be traced to criteria written before the period began and to evidence recorded during it, and the employee gives their own account before hearing the manager's. Afterwards, someone checks the spread of ratings across groups. Performance review bias rarely shows up in a single review; it shows up in the pattern across many, which is why the check matters more than any one manager's good intentions.

Calibration meetings, where managers explain their ratings to each other against the same criteria, catch some of it. A 360 review spreads the judgement across more people, which thins individual bias without removing it; the 360 performance review template keeps raters anonymous in groups of at least three for that reason.

Where training fits

Awareness training on its own has a weak record at changing behaviour. Frank Dobbin and Alexandra Kalev's review of corporate diversity programmes found that mandatory training rarely improved outcomes and sometimes made them worse. That is unsurprising: knowing that affinity bias exists does not tell a manager how to write a criterion that resists it. What changes behaviour is being taught the mechanics, how to write an objective that can be assessed on evidence, how to keep a record that survives challenge, how to run a conversation that does not depend on the manager and employee sharing a background. UCSF's unconscious bias resources are a useful companion to that, rather than a substitute for it.