HR Has a Performance Review Problem. And It's Not Their Fault

Imagine it's review season. HR has sent out the templates, managers have their deadlines, and everyone is supposed to come prepared with examples.

Then calibration starts.

One manager can explain exactly why she thinks an engineer is performing above level. Another says, "I just don't think he's quite there yet." When someone asks for examples, the answer is a few recent projects and whatever the manager remembers from the last couple of months.

Neither manager is necessarily doing a bad job. They simply have different amounts of evidence to work with.

This is one of the harder problems with engineering performance reviews: HR can make the process consistent, but it can't make the information going into it consistent.

The Evidence Problem

Engineering makes this particularly difficult because important work doesn't always appear neatly in a project tracker.

An engineer might spend several days helping another team solve a technical problem, take ownership of an incident, mentor someone, or prevent a larger problem from happening. Some of that may show up in Jira or GitHub. A lot of it won't.

By the time the review arrives, the manager is trying to reconstruct the story.

That's where vague statements like "great problem solver" or "strong performer" become a problem. They might be true, but they don't tell HR, the employee, or another manager what actually happened.

A better example would be:

"In March, Sarah took ownership of the payment migration after the project lead left. She coordinated with infrastructure, found two issues in the rollout plan, and got the migration completed without delaying the launch."

Now there's something everyone can discuss.

Calibration Can't Create Missing Evidence

Calibration meetings can help managers apply standards more consistently, but they can't recreate information that was never captured.

Recency bias makes this worse. A mistake from last month is easier to remember than something important that happened eight months ago. The same thing happens when someone has a few impressive wins just before review season.

The answer isn't documenting everything an engineer does.

It's capturing the moments that will matter later.

What HR Can Change

A simple improvement is to give managers a clearer standard for evidence.

For important assessments, ask:

• What happened?
• What did the person do?
• What changed because of it?
• What context matters?

That doesn't require another lengthy form. A few sentences can be enough.

HR can also look at the evidence behind ratings during calibration. If one manager rates almost everyone highly while another rarely does, don't immediately assume one is wrong. Look at how they're assessing performance and whether they're using similar standards.

Don't Wait Until Review Season

The easiest time to remember why something mattered is when it happens.

If someone handles a difficult incident, receives meaningful feedback, takes ownership of a struggling project, or repeatedly misses an expectation, that's when the manager should capture enough context to remember it later.

The goal isn't to create a surveillance log.

It's to make sure that when review season arrives, managers aren't trying to remember six months of work from scratch.

A good performance process should make decisions easier to explain, not just easier to administer.