The review form I worked with last cycle had five freeform text boxes, a minimum of 250 words each, asking me (the manager) to describe how the engineer demonstrated collaboration, ownership, technical depth, communication, and growth. The review form came with guidance and instructions, and mentioned that this would all be reviewed and cross-referenced as part of a “calibration” meeting later. What struck me was the subjectivity of it all; and why any of this was necessary.
Nobody writes an essay to determine whether the website is performing optimally. You look at the metrics, the metrics are the answer, and there’s nothing left to discuss unless somebody wants to know why. Think about how you judge the performance of your barber or hair stylist. It’s not that nuanced. They either consistently provided the services you asked for, or they didn’t. You continue to patronize them on that basis. That’s how you evaluate a person’s performance in your personal life, and it can (and should) work the same way in business as well. Once leadership defines what it expects and what a good result looks like, and the responsible party commits, then whether a person delivered stops being a matter of opinion. Requiring narrative is an admission that nobody defined a job well done.
Why Calibration Exists
Start with the calibration meeting, because it’s very telling in and of itself. Two managers show up with two engineers, and two narratives, to compare. One of them has written tight, specific paragraphs with examples you can picture. The other writes four sentences with ambiguous language that amounts to a “meets expectations” rating. The first engineer is perceived as the higher performer, and everybody leaves the room believing they made a careful call about performance.
They made a careful call about writing, that’s it, and if you picked up on the fact that having a bad writer as your advocate could affect your performance review, you’re not crazy, that’s entirely possible (even probable).
Calibration exists to correct for bias and subjectivity, and approximate objectivity via consensus. It’s all based on the flawed premise that objectivity is not possible. If leadership leaned into objectivity there would be nothing to calibrate, and the meeting would be appropriately about alignment.
What a Performance Review Is For
A performance review exists to provide feedback, to the individuals and the organization, about the individual’s performance, primarily so that people can make pay and promotion decisions even if they had no direct involvement with the person being reviewed (or their work). That’s its function. Nobody enjoys the review process, neither reviewer nor reviewee.
Every broken review process I’ve seen (and been a part of) was most faulty due to its subjectivity. Reviewees are asked to recall everything they’ve done, that they would like to be considered (on the record), that they would like to get “credit” for. If you’re thinking: “wouldn’t the people who commissioned the work already know what was done and whether it was a success or failure?” you’re asking disruptive questions and outing yourself as difficult and “not a team player”.
Lines of Code, and Other Metrics
An alternative in the direction of objectivity is swapping the narrative for a dashboard. The dashboard shows tickets closed, commits merged, lines of codee written, story points burned down, etc. This is almost universally loathed, especially in the tech industry.
Counting tickets is measurement. It’s accurate, it’s checkable, but it’s aimed at the wrong thing. It’s objective, sure, but it invites gamification and misdirection.
A ticket count tells you what somebody did. It doesn’t tell you whether what they did produced the desired outcome. It has the same defect as the narrative-based review, i.e., it’s focused on recalling activity, which is why trading one for the other buys you nothing. Instead the focus should be on affirming and reporting on commitments and outcomes.
Outcomes and Objectivity
Is it possible to use a single metric by which you can evaluate a person’s performance? Yes, and that metric is outcomes, but it requires discipline all around, not just for the operator. Outcome-based reviews are simple. If we’ve done the upfront work of aligning on expectations and desired outcomes, then the performance evaluation becomes objective. The outcome was either realized or it wasn’t.
Common Objections
The objection I hear most is: what happens when a project slips and it isn’t anyone’s fault. A dependency collapsed, the date moved, none of it was the fault of the operator. Doesn’t that need a clarifying paragraph during a review?
My take is “No.” It needed clarification and correction when it was first discovered that the project was at-risk.
A commitment is a live thing. When the ground moves under it, whoever owns it renegotiates right then, with the people counting on it, and writes down what replaced it. Do that and the record shows a commitment that changed and a result that landed against the new one. Skip it and the record shows a miss, because a miss is what it was. If the first anybody hears about a slipping date is at review time, the silence is the failure, and no paragraph after the fact repairs it.
Etiquette/Conduct is not Performance
The other case (that I hear as an objection) is the engineer who hits every number but disrupts the culture. That’s a real problem, but it doesn’t belong in the performance review. Conduct gets handled when it happens, on its own timeline, with its own consequences. Holding onto an issue for eleven months so it can be laundered into a rating about performance is how you get “sometimes abrasive in group settings” as part of a conversation about how good you are at your job.
Where the Narrative Goes
None of this is meant to suggest that the narrative has no place. It simply needs to be demoted to the role of supporting information around an otherwise objective measure of a person’s efficacy at work.
