River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Performance Review Process Template

Five documents and three sheets, including a tracker that projects where the cycle closes by weighting every outstanding review at its own manager's prior rate.

Free download  ·  No account needed

Rating Distribution, one working day

Two ways to read a half-finished cycle, and only one of them holds

One row per working day the window is open. The last row is the cycle. Every row before it is a forecast, and the sheet is explicit about which is which.

Submitted And Outstanding

Counts, not percentages. On day one this pair is the only honest thing on the row, and the sheet says so in the note column rather than leaving somebody to work it out.

Running Share Above The Middle

The share of what has arrived that sits away from the middle and above it. This is the number every review platform shows you and the number that will mislead you, because a manager writing an easy review submits it on Tuesday and a manager writing a hard one submits it on the last afternoon.

Projected Close, Each Manager's Own Prior Rate

Every outstanding review weighted at the rate its own manager produced last cycle, added to what has already arrived. This is the column to plan against, and the per-manager part is not a refinement. A manager who put nobody above the middle last time contributes nobody this time.

Projected Close, Company Prior Rate

The same arithmetic with one company average standing in for every manager. It is here as a control, so the sheet can show what the shortcut costs rather than assert it.

Evidence Returns Issued, Cumulative

The one mid-window figure anybody should act on. It is about how an evidence box was filled in, it belongs to one named manager, and it decays: a return issued on day nine cannot be answered before the window shuts.

Six working days into a ten-day submission window, 74 of 191 reviews were in and 66.2 percent of them sat above the middle of the scale. The cycle closed at 51.3 percent, and nobody did anything in between. Submissions arrive sorted by how comfortable the review was to write, so the front of every window is the top of every scale. Reviews that arrive already specific trigger less of everything downstream, which is what review writing is for.

The per-manager weighting is the whole trick. Run the same projection at the company's prior rate and day 6 reads 57.3 percent, six points high, peaking exactly when somebody decides whether to intervene. One Marloe Logistics manager held fourteen reviews, none of them above the middle last cycle, and filed all fourteen on the final day. Weighted at his own zero rather than the company's 52 percent, the projection stayed inside a 2.6-point band all window, which is stable enough to plan against.

So the projection mostly exists to stop an intervention. A mid-window note saying ratings are running high tells managers the answer before they write the hard reviews. A rating steered toward a target cannot be explained one at a time afterwards, and a procedure deciding a promotion or a retention decision has to be. What is correctable instead is the evidence, one review at a time. Eleven rows reached calibration against 31 last cycle.

The running share said 66.2 percent on day 6 and the cycle closed at 51.3

The Rating Distribution, the Cycle Tracker by Manager behind it, and the Completion Status at close.

Rating Distribution

Illustrative for a fictional warehouse software company, Marloe Logistics. H1 FY27, 191 reviews across 25 managers, window open ten working days.

DayDateInOutAbove
middle
Running
share
Projected,
own rate
Projected,
company rate
Returns
16 Jul4187375.0%49.4%52.1%2
27 Jul11180872.7%49.6%52.8%5
38 Jul221691463.6%49.1%53.0%9
49 Jul381532565.8%48.7%54.4%15
510 Jul571343866.7%49.4%56.1%17
613 Jul741174966.2%50.8%57.3%19
714 Jul96956062.5%51.0%57.1%24
815 Jul124677258.1%50.7%55.8%29
916 Jul158338855.7%51.0%55.0%35
1017 Jul19109851.3%51.3%51.3%41
Prior cycle H2 FY26 for reference: 186 reviews, 96 above the middle, 51.6 percent, and 31 rows reached the calibration room.

Final spread 16 Outstanding, 82 Exceeds, 85 Meets, 8 Below. The running share sat above 62 percent for seven of the ten days and was 14.9 points high at the midpoint. The projection's whole range across the window was 48.7 to 51.3 percent. The company-rate control peaks on day 6, which is the day somebody would have acted on it.

Cycle Tracker by Manager

Sample rows from a 25-manager view, as at close of business on day 6. The prior rate column is what the projection reads.

ManagerTeamReviewsPrior rateIn by
day 6
Above
middle
OutProjected
from those
Returns
issued
Last review
arrived
Priya RamanImplementation1182%11800.06day 4
Gregor LindqvistSales967%9600.04day 5
Rosalind AchebeSolutions862%8500.04day 3
Dominic SearleEngineering944%5341.80day 7
Joanne FerrisPeople Ops1050%00105.00day 7
Hester LindqvistImplementation838%0083.00day 8
Grace OyelowoMarketing650%0063.00day 10
Owen WhitlockCustomer Support140%00140.00day 10
Company19152%744911748.119day 10

Owen is the row that decides the whole projection. Fourteen reviews, none above the middle last cycle, all fourteen filed on the final afternoon. At the company's 52 percent his team contributes 7.3 above-middle ratings that were never going to exist. At his own rate it contributes none, and that single substitution is most of the six-point gap between the two projection columns. Of the 19 returns issued by day 6, 14 belong to Priya, Gregor and Rosalind.

Completion Status

One row per check rather than per person, so the cycle can say what it did about each and who closed it.

CheckCountOfClosedNote
Reviews in scope191240 staff17 JulFirst 90 days excluded, and given a 90-day conversation instead.
Ratings away from the middle10619117 JulThe only ratings the evidence check reads.
Evidence returns issued4110616 JulNo date, no artifact, no figure, no third party.
  Came back checkable304117 JulA date, an artifact, a figure or a named outsider.
  Manager moved it themselves84117 JulNobody asked them to. They reread their own words.
  Stood as written34117 JulLegitimate. Goes to calibration with both versions.
Goals in force at rating time1911916 JulRegister frozen the morning the window opened.
Rows reaching calibration1119120 Jul3 evidence gaps, 5 two-level moves, 3 contradicted, 0 retired goals.
Ratified in one act18019120 JulNamed signer, dated, against the criteria in the plan.
Acknowledgment recorded18919114 AugTwo on leave, both dated for their return.
Retention floor1 Sep 2027n/aOne year from the pay change, which falls after the record date.

The zero on the retired-goals row was six last cycle, and the freeze is the entire reason. Those six were never rating disagreements: an objective had been retired in October when the roadmap changed and the manager rating against it had not been told, so re-rating them was clerical work that had been dressed up as a judgement call and given meeting time.

What is in the pack

01

Rating Distribution

One row per working day: what is in, the running share, the projection at each manager's own prior rate, and the same projection at one company average sitting beside it as a control.

02

Cycle Tracker by Manager

Reviews owed, prior rate, what has arrived, what each manager's outstanding batch is projected to contribute, and their evidence returns issued and still open.

03

Completion Status

One row per check with a count, an owner and a closing date, including the acknowledgments and the retention floor, which is one year from the record or the personnel action, whichever falls later.

04

Cycle Plan

Scope as a number with a rule behind it, dates counted in working days, and the decisions assigned rather than the tasks. It says in as many words that the cycle owner owns no rating.

05

Review Forms by Level

Four forms on one spine, differing only in what counts as evidence at that level, with the goals pre-filled from the frozen register so a retired objective cannot be rated against.

06

Manager Writing Guide

Two pages of before-and-after rewrites for the four constructions that get a review returned: a trait, a bare number, an absence of problems, and a comparison to a colleague.

07

Calibration Agenda

The eleven open rows with the evidence text quoted as written both times, the ratification block with its signer, and the structural findings addressed to the next cycle rather than the room.

08

Why the Running Distribution Is Not the Cycle

The two pages to read before the tracker, on why early submissions are sorted by rating and what the per-manager weighting is actually correcting for.

How it works

  1. 1

    Send last cycle's ratings

    Per person, with the manager who wrote each one. Plus whatever holds your goals now and your headcount with manager and level. Inconsistency between teams is expected and is not worth tidying first.

  2. 2

    The goal register gets frozen

    Every goal gets an effective-from date and a retired-on date where one applies, before the window opens. From then the form offers a manager only the goals in force for that person, pulled from the same cascade map that flagged which ones connect to a stated objective.

  3. 3

    The window runs and reports honestly

    Each arriving review is checked the day it lands, and the projection reads the close from whatever has arrived. When it holds within a point of last cycle, the finding is that nothing needs doing.

  4. 4

    Calibration gets a closed set

    The rows still open, each with its reason and its evidence quoted as written, plus a ratification block with a named signer for everything else. The projection stays behind.

Frequently asked questions

Why not just read the distribution as reviews come in?

Because the order they arrive in is not random. An easy review takes twenty minutes and gets written on Tuesday; a hard one needs a prior record found and a sentence composed, and it slides to the last afternoon. In the worked cycle that read 66.2 percent above the middle on day 6 against a 51.3 percent close.

Why weight by each manager rather than one company rate?

Because the managers are not interchangeable, which is what your last calibration probably found. One manager here had fourteen reviews and put nobody above the middle last cycle. At the company's 52 percent his outstanding batch contributes 7.3 ratings that never appear. His own rate is zero, and using it is what makes the projection work.

What if this is our first cycle and there is no prior data?

Then the projection has nothing to weight with, and the pack says so rather than substituting a company average and presenting it as a forecast. Run the evidence check and the goal freeze this time, which need no history, and the tracker records each manager's rate as it closes so next cycle has one.

Does returning a review to a manager mean overriding their rating?

No, and the difference is the point. The check reads the evidence box and ignores the rating's level entirely. It quotes the sentence and asks for a date, an artifact, a figure or a named third party. Eight of 41 managers moved the rating themselves on rereading; nobody asked them to and the cycle owner rates nobody.

How is this different from the calibration template?

This one runs the cycle and hands over a closed set of rows. The calibration pack computes the session agenda from submitted reviews and finds the manager patterns in them. Eleven rows reached the room here against 31 last cycle, which is what the evidence check and the goal freeze bought.

What format are the downloaded files?

Five Word documents and three CSV sheets in one zip, needing no conversion. The documents open in Word, Pages and Google Docs and the sheets in Excel, Numbers and Google Sheets. A PDF version is there for circulating the plan and the writing guide rather than filling them in.

Is it free, and what does Edit with AI add?

The download is free with no account. Edit with AI installs the same pack as a private Space and runs it against your own material: your prior ratings, your goal records, your headcount. It freezes the register, projects your close, and checks the reviews as they arrive.

Find out where this cycle actually closes

Send last cycle's ratings with the manager who wrote each one, plus whatever holds your goals now. The first thing back is the projected close from what has already arrived, against last cycle's, and the reviews that owe you an evidence sentence.

Edit with AI