Performance Management

Companies judged people long before the term existed. The judging just happened once a year, all of it crammed into December. A manager would picture each report in turn, assign a rating, and hand the ratings to human resources, which turned them into next year's raises.

That arrangement had two holes in it. First, memory doesn't cover a year evenly. The outage someone spent a weekend containing in March fades; one wrong statement in a November meeting stays sharp. Ratings ended up measuring the last two months rather than the last twelve. Second, the person being rated heard the outcome without hearing the standard behind it. Without the standard there is no way to know what to do differently.

The year-end conversation followed a predictable script. The manager said the rating was a B, the report asked why, and the manager said that taking everything together, that's where it landed. The report left unconvinced, and the manager, with no record to point at, disliked the meeting as much as the report did. A single ten-minute exchange carried a whole year of judgment plus salary plus prospects, so both sides arrived defensive. What came out of it regularly turned up as the top reason people gave for leaving the following spring. The person rated checked the raise and stopped listening; the person rating remembered it as the worst hour of their year.

Performance management pulls that judgment out of one December afternoon and spreads it across the year. The standard goes up front, the checking happens continuously, and the end of the period is for collecting what has already accumulated. The front half of that work — deciding what to achieve and writing down the numbers that will confirm it — belongs to goal setting, where splitting an aim into an objective and its key results, known as OKR, is the most common approach. Performance management takes the back half: comparing results against the standard and adjusting what comes next. Evaluate without goals and the evaluator's impression becomes the standard; set goals without evaluation and the quarter ends with a document nobody reopens.

The terms under performance management fall into four groups: four formal procedures that run on a cycle, two everyday conversations that fill the gaps between them, one response for when expectations aren't being met, and one stance that makes all of it work.

Before the groups, it helps to see how one turn of the cycle goes. Most companies run it on a half-year or quarterly rhythm, in four steps.

1. At the start of the period, manager and report agree on what this period is for and write it down. Skip this and the other three steps lose their footing.
2. During the period, short conversations repeat. A thirty-minute one-to-one every other week, plus comments on what went well or sideways as it happens. Most of the cycle's actual time lives here.
3. When the period closes, everything that accumulated gets written up once. This step collects conversations that already happened rather than producing new verdicts.
4. Several evaluators review the write-ups together to align their standards, the result goes back to the person, and it feeds the next period's goals, along with decisions about pay and level that hang off them.

The procedure group makes up steps 3 and 4.

1. Performance review — the procedure that turns a period's results and capability into a judgment. A self-assessment written by the person alongside an assessment written by the manager is the usual format. This is the center of the group; the other three are inputs to it or cleanup after it.
2. Peer review — assessment by the people who worked alongside someone, added to cover what a manager can't see. Whether an engineer handled another team's requests gracefully is known only to whoever sat next to that work. Reviewers are usually nominated by the person, three or four of them, or assigned from records of who actually collaborated. Letting people nominate freely tends to produce a list of friends, so managers check that the names reflect the quarter's real work.
3. 360 feedback — input gathered from several directions at once: peers, the manager, direct reports, sometimes partner teams. Including the manager as a subject is what separates it from peer review. When five of six team members write down the same thing, that isn't one person's grievance, it's a pattern in how the team is run.
4. Calibration — a meeting where evaluators sit together and align their scales. One manager gives most of the team the top rating and another gives it to nobody; left alone, two people doing comparable work get different outcomes depending on whose team they sat on. Five team leads spending two or three hours on it is typical, and a rating nobody can justify comes down in the room. The point isn't to force a distribution but to make each evaluator say out loud what a given rating required this period.

The everyday group fills step 2, and it's where the cycle's outcome is effectively decided.

1. One-on-one — the regular private conversation between a manager and a report, most often thirty minutes every other week. It's for what's blocked and where things are heading, not for status reporting. Having the report set the agenda keeps it that way; a manager arriving with a list of questions turns it into a weekly status meeting within a month.
2. Feedback — telling someone what worked and what didn't, close to when it happened. Held feedback loses value fast. A meeting that ran long because someone talked past the agenda changes the next meeting if it's mentioned that day, and reads as an ambush if it first appears in a review four months later.

The response group has one term. A performance improvement plan (PIP) writes down what has to change, by when, and to what level, and concentrates support on that window, usually thirty to ninety days. Many people read it as a firing procedure, though it was designed for the opposite. When shortfalls have only ever been mentioned out loud, neither side has ever agreed on what different would look like, and this is the first place that standard gets written down. The document carries the expected level, how it will be checked, and the deadline, along with the manager's own commitments to check in weekly and supply what's missing. At the deadline, the only test is what the document says.

The stance group also has one term. Radical candor means caring about the person and being direct at the same time. Where the other three groups are procedure, this is the posture that drives them. Soften everything to protect the relationship and the other person never learns a problem existed; be blunt without care and they remember the delivery instead of the content. The hard part is that the two failures look like opposites, so a manager correcting one usually overshoots into the other for a while. However tight the procedure, without this posture every sentence in the write-up comes out bland.

The four groups connect in one direction. One-on-ones and feedback build the record during the period, the performance review organizes that record, peer review and 360 feedback add what the manager couldn't see, and calibration aligns the scales between evaluators. If someone still falls short after all of that, the improvement plan takes over. Skipping steps usually shows up at the end. Say nothing during the period and raise a problem for the first time in the write-up, and that document stops being a judgment and becomes a notification. When the period had enough conversation, the last step gets short, because all that remains is confirming in writing what both people already know.

"Nothing in this review should surprise you. We covered all of it in our one-on-ones."

The end-of-period document is a summary of conversations already held, not news.

"That rating moved in calibration. The teams were using different scales."

One manager's judgment doesn't stand as the final answer; the evaluators met and aligned.

"We never wrote the goals down in week one, so the evidence for this review is thin."

Grading results with no standard in front of them means falling back on impressions.

Pagedeck put a real performance cycle in place the year it passed forty-five people. Until then the whole thing was Grace running a two-hour meeting with team leads each December to settle raises. Two people left back to back that spring, and both said the same thing in their exit conversations: they had never once been told how they were doing.

Grace changed three things and started running half-year cycles. Managers hold a thirty-minute one-to-one with each report every other week, goals go in writing and stay visible from the start of the half, and the write-up at the end compares results only against those goals. The first cycle was rough. Mia spent five hours a month on conversations with her five reports, and Theo took three days just to phrase his goals. The end of the half looked different anyway. No review meeting contained a first-time criticism, and the meetings ran half as long, because they were confirming something both people already knew. Grace kept the same three rules the following half and added nothing to them.

The most common misreading is that performance management is paperwork that happens during review season. It looks that way because the only point most people touch it is an email telling them to fill in a form. Break down the hours in one cycle, though, and the write-up is maybe a tenth of it; the rest is the private conversations during the period and the comments made as things happen. Leave that order inverted and there is nothing to write from at the end. With no record to lean on, people lean on recent memory, which puts the evaluation right back where it was before the system existed. This is how a company ends up with more procedure and identical results. When reports complain about the same things they complained about before, counting how many one-on-ones actually happened last quarter gets to the cause faster than redesigning the form.

The second misreading is that the system exists to screen out low performers. Improvement plans and bottom ratings are the visible cases, and most people have heard at least one story where those led to an exit. In practice, nearly everyone going through a cycle is performing as expected, and what they get out of it is not a grade but a direction for the next half and a sense of where they're growing. Read as a screening device, it turns every review into a defense. The way to win a defense is to commit only to what you're sure of, so the goals submitted at the start of each period get safer and safer. Goal setting collapses first, and performance management, sitting on top of it, goes slack with it. Whether someone who aimed high and reached half of it ranks above someone who aimed low and reached all of it is the question that settles which way this misreading goes.