Performance Review

At ten people, the review happened inside the founder's head. Someone set aside a day in December, thought about each person in turn, and wrote down a raise percentage. With no record to work from, memory was the evidence, and memory tilts toward the last month and a handful of vivid events.

So two people who worked the same half came out with results nobody could explain. At Pagedeck, the engineer who stayed up fixing an outage two weeks before the review was remembered clearly. The engineer who spent the whole half making sure outages didn't happen went unmentioned. One of those is an event and sticks; the other is the absence of events and leaves nothing to picture.

The problem ran in the other direction too. The December conversation was where people first heard about a problem with their work. Nobody had said anything for six months, and then it arrived as the reason a raise was small — a verdict delivered after the chance to fix anything had passed.

A performance review is the procedure built to block both of those, and it does it by fixing the standard before the work starts rather than after it finishes. Write down what the period is for while the period is starting, then put the results next to that document when it ends. Attaching a rating matters less than the fact that the standard was written in advance.

Six steps. The first half accumulates during the period; the second half lands in the three or four weeks after it closes.

1. At the start of the period, manager and employee agree on what this cycle is meant to produce and write it down. Three to five goals stay manageable. Write ten and nobody remembers by the end which ones mattered.
2. Through the period, progress gets checked in one-on-ones — the recurring private conversation between manager and employee, commonly 30 minutes every other week. Whatever those conversations leave behind becomes the raw material for the write-up.
3. When the period closes, the employee writes a self-review that walks through the goals from step 1 with results and evidence. Not "made good progress" but "the target was 4 percent free-to-paid conversion, it landed at 3.6 percent, and here is why."
4. Peer reviews get collected — three or four colleagues describing what working together was like. Widen the same exercise to include managers and reports and it becomes 360 feedback, where the manager is also on the receiving end.
5. The manager writes the evaluation, setting the self-review, the peer input and the one-on-one notes side by side in one document. The rule is that no new judgment gets created here. A criticism that appears in the write-up without ever having been said during the period has no evidence behind it.
6. Reviewers meet to line their standards up, the result goes to the person in a conversation, and it feeds the next cycle's goals. That alignment meeting is calibration.

Four kinds of people touch one person's review. The employee writes a self-review, three or four colleagues write peer reviews, the direct manager writes the evaluation, and managers from other teams look at the rating together in calibration. Skip that last seat and standards drift team by team, so two people who worked at the same level get different outcomes depending on who they report to.

Half-yearly is the most common cadence. Quarterly means judging work before the results exist; annually means the first six months fall out of memory. At 45 people a full cycle takes three or four weeks, and during those weeks roughly half of every manager's working time goes into this paperwork.

The timing gets counted backward from the pay calendar. If salary changes take effect in April, ratings have to be final by mid-March, which puts the start of self-reviews in mid-February. Promotion decisions usually ride along in the same round. A promotion moves someone to a higher level or role, and most companies decide it off the review results rather than running a separate process for it.

One part of the order never moves: the goals get agreed at the start. Write them after the period ends and they get written to match the results already in hand, which turns the review from a judgment into a transcription exercise.

"This round is written against the goal document we agreed in January. Anything outside it, note separately."

Fixing what counts as evidence before anyone starts writing.

"Don't just put results in your self-review. Attach the numbers and the document links."

A request for claims in a form the manager can actually verify.

"This is the first I'm hearing of that, and it's in my review."

Nobody raised it during the period, so the judgment is hard to accept — and the complaint is usually about the silence rather than the rating.

Pagedeck runs the cycle every half year. Last round Mia hit three of four goals; the fourth finished at 3.6 percent against a target of 4. Her self-review led with the one she missed and attached the record showing two schedule changes behind it.

When Grace wrote the evaluation, that document wasn't her only source. She worked from four: the biweekly one-on-one notes from across the half, what five colleagues had written, and the goal document the two of them had drafted at the start. Nothing new came up in the conversation. Confirming what they both already knew took half an hour, which is roughly how long the conversation runs when the period was documented properly.

The most common reading is that a performance review is the place where ratings get decided. A rating does come out the end, and it does attach to pay, so the assumption is reasonable. But what actually produces the result is the record built during the period. Write an evaluation at the end of a half with no one-on-ones and no goal document, and the manager ends up summarizing six months from two or three remembered scenes. However carefully the process is designed, at that point the outcome matches the era before there was a process.

The other mistake is believing a well-written self-review changes the outcome. Polished sentences feel like they should help, but what the reader is looking for is verifiable evidence rather than phrasing. Three lines with a number and a document link behind them beat a full page of adjectives, and without evidence even a well-written claim stalls the moment another manager asks about it in calibration.