My manager asked me to rank my coworkers in a private spreadsheet

My manager sent me a spreadsheet titled “Team calibration input.”

It listed eight coworkers and asked me to score each person from one to five on reliability, judgment, communication, and growth potential. A final column asked who I would trust with “higher-stakes work.”

I am not a manager. These people are my peers.

The message said my responses would remain confidential and help leadership prepare for annual reviews. My manager described the exercise as a way to gather perspectives that formal reporting lines miss.

I can see the appeal. I work closely with people my manager observes only in status meetings. I know who catches errors, who shares context, and who disappears when work becomes difficult. Peer input can reveal important contributions.

But the spreadsheet does not ask for examples. It asks for rankings.

Two coworkers do very different jobs. One handles a steady operational queue; another works on ambiguous projects with fewer visible outputs. Rating both on the same five-point scale feels precise without being fair. “Growth potential” is even harder. I have no idea what opportunities they have requested or what feedback they have received.

The confidentiality worries me too. If these scores affect reviews, the people being rated may never know which impression lowered them or have a chance to correct incomplete information. I would not want my own career reduced to anonymous numbers from coworkers who see only part of my work.

I asked my manager how the data would be used. She said it was one input among many and told me not to overthink it. I asked whether I could provide examples without numerical rankings. She said the spreadsheet needed to remain consistent across reviewers.

I do not want to refuse all peer feedback. I am willing to describe work I directly observed: who contributed what, what happened, and where collaboration broke down. I am not comfortable assigning a hidden potential score.

My current plan is to return the sheet with only evidence-based comments and mark the rating fields “not enough information” where that is true. I will also state that I cannot assess growth potential for peers I do not supervise.

Would you complete the rankings, narrow the response, or decline the exercise entirely? What would a fair peer-feedback process ask that this spreadsheet does not?