River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Compensation Review Process Template

Four documents and four sheets that total every manager's merit allocation and show which pay differentials the cycle widened, before it is approved.

Free download  ·  No account needed

A merit cycle is 24 managers allocating a budget across their own teams, and each one can be completely defensible while the total widens a differential nobody proposed. Approval happens manager by manager too, so the person signing never sees the artifact that causes it. This pack runs the equity check across the whole proposal, once, before anything is approved. In the worked example that surfaced four job groups and eleven people, all of them invisible from inside any single team.

The second mechanism is the unit of measurement. Every recommendation is converted into a change in position in band, not a percentage, because an equal 3 percent on unequal bases widens the money gap and leaves relative position where it started. An equal allocation is a regressive allocation, and the percentage view cannot show it: both numbers on the screen read 3.0. The ratings feeding the cycle should have survived a calibration session first, or the cycle is spending money on unequal scales.

Third, the correction budget comes out of the pool before managers allocate anything. Marloe Logistics reserved 0.4 of 3.2 points up front, which funded all eleven corrections inside the same approval rounds and cost the merit pool four tenths of a point. Find the problem after the money is committed and it becomes an off-cycle request in Q3, which gets declined. Written for people leads and finance partners who have signed a cycle they could not later explain. The reviews behind the ratings come from review writing.

The number no cycle computes

Three numbers per job group, in this order: before the cycle, if approved as submitted, after corrections. The middle one changes decisions and requires every manager's submission in one sheet before any of it is approved, which is why a normal cycle never has it.

Budget Allocation Model

Illustrative. Marloe Logistics, fictional 240-person warehouse software company. FY26 cycle, 203 eligible, effective 1 April.

LineAmountPointsNote
Eligible payroll20,320,000203 of 240, employed on or before 1 October
Board-approved budget650,2403.20The only number finance tracks
Less correction reserve-81,280-0.40Taken BEFORE manager pools were calculated
Merit pool to managers568,9602.80What 24 managers were told, and why
Merit allocated568,9602.80203 awards, 41 of them zero
Corrections released67,0320.3311 people, average 5.9% of post-merit base
Reserve returned unspent14,2480.07Back to finance, not redistributed as merit
Total cycle spend635,9923.13Against 3.20 approved

The reserve cost every manager pool 14.3 percent of what it would otherwise have been. Nobody noticed, because nobody knew what the pool would otherwise have been. That is the whole trick: a correction found after allocation closes has no money behind it, so it becomes an off-cycle request in Q3, and Q3 off-cycle requests get declined.

Equity Check

Grouped by job code and level, never by manager. 12 of 31 groups passed a stated population threshold of 6, covering 114 of 203 eligible employees.

Job groupNGap beforeGap as submittedGap afterCorr.Cost
Implementation Consultant L31413.014.110.9324,447
Data Analyst L2812.714.110.1212,548
Account Manager L3911.613.29.7315,677
Support Engineer L21110.711.98.9314,360
Software Engineer L31310.210.010.00
Seven further groups59all narrow0
19 groups below threshold89Not checked. Count reported, no ratio computed, all 19 named so next cycle does not assume they were cleared.

Gap = difference in average position in band between the below-midpoint and above-midpoint halves of the group. Implementation Consultant L3: six below averaging 91.0 percent of midpoint, eight above averaging 104.0. Proposed awards 2.9 and 3.6 percent, both tracking ratings, nobody out of policy. As submitted that is 93.6 against 107.7.

Corrections are upward only, so 107.7 does not move between the second and third columns. A gap that closes because somebody above the midpoint got less is a cut with a better name.

Recommendation by Employee

Implementation Consultant L3, 14 people across three managers. The group crosses three reporting lines, so it appears on no manager's screen during a normal cycle.

EmployeeMgrPos. beforeRatingAwardPos. after meritToward/awayCorrectionFinal
Nadia ColomboFerreira87.0Meets2.6%89.3Away+11,35998.0
Femi AdeyemiFerreira89.0Exceeds3.7%92.3Toward+7,41998.0
Priya RamanBerg91.0Exceeds2.9%93.6Away+5,66998.0
Sofia LindqvistPark92.0Meets2.7%94.5Awaydeclined94.5
Hakim SarrFerreira93.0Meets2.6%95.4Away95.4
Rosa VillalobosBerg94.0Meets2.9%96.7Away96.7
Ines BakkerPark100.5Exceeds3.6%104.1Toward104.1
Owen CastellanosBerg101.5Meets2.9%104.4Away104.4
Meera AnandFerreira103.0Exceeds3.6%106.7Toward106.7
Tobias WeissPark104.0Outstanding4.4%108.6Toward108.6
Lucia MorenoFerreira105.0Exceeds3.6%108.8Toward108.8
Daniel OkaforBerg106.0Meets2.6%108.8Away108.8
Yuki TanakaPark107.0Outstanding4.2%111.5Toward111.5
Marcus OyelaranFerreira105.0Exceeds3.6%108.8Toward108.8
Below midpoint691.02.9%93.624,44796.8
Above midpoint8104.03.6%107.7107.7

Toward or away compares each person's move in band position against the move of the group median, which is the only view of the aggregate a manager gets from inside their own team. Nine of fourteen rows are individually defensible and move away.

Sofia Lindqvist's correction was declined and recorded as declined: 1.8 years in role against a group median of 2.8. A declined correction that nobody writes down gets rediscovered as a new finding next April.

What is in the pack

01

Budget Allocation Model

The pool, the correction reserve taken out of it before manager pools are calculated, and what that reserve costs the merit pool in points.

02

Recommendation by Employee

Every eligible person with base, midpoint, position in band, rating, tenure in role and proposed award, plus band position before and after.

03

Equity Check

Three numbers per job group: before, if approved as submitted, after corrections. Cut by job code and level, plus aggregate demographic cuts at company and family level.

04

Approval Tracker

One row per round with a named approver and what changed as a result, including declined corrections and conditions attached to an approval.

05

Cycle Guidance

Three pages that go to managers before allocation opens. The pool, eligibility, what they are allocating against, and what will bounce.

06

Manager Brief and Communication Templates

What each manager is told after the check, and the four scripts they use: standard award, correction, zero, and below target.

07

Two rules every prompt reads first

The cycle is checked as one proposal, never manager by manager. Every zero award needs a written reason on the row.

How it works

  1. 1

    Send payroll, bands and ratings

    Current base by person, the band structure with midpoints, performance ratings, the budget, and location and tenure if you hold them.

  2. 2

    Get the position-in-band baseline

    Every eligible person converted from a salary into a position against their own band midpoint, before anybody allocates a single point.

  3. 3

    Allocate against a reserved pool

    The correction budget comes out first. Managers see a toward-or-away column beside each award, which is the only aggregate view available from inside a team.

  4. 4

    Run the check on everything at once

    Three numbers per job group before any approval, corrections priced to each comparator floor, and a tracker recording who approved what and what changed.

Frequently asked questions

Why position in band instead of percent?

Because an equal percentage on unequal bases is a regressive allocation. Two Implementation Consultants at 91.0 and 104.0 percent of midpoint who both get 3 percent finish 507 dollars further apart and in almost the same relative position. The percentage view shows 3.0 and 3.0 and cannot show anything else. The midpoints come from the band structure.

Does this override my managers' judgement?

No award was reduced and no rating was questioned. The four flagged groups got corrections funded from a separate reserve, upward only. Three of 24 submissions came back, all for a blank reason on a zero award rather than a number anyone disputed. Every brief says which pot funded what.

Is this a pay equity audit?

It is the in-cycle version: does this proposal, taken together, widen something no individual recommendation created. Federal contractors must evaluate compensation systems for gender, race and ethnicity disparities (41 CFR 60-2.17). A full study looks backwards and is a pay equity analysis. This looks at money not yet committed.

What happens to job groups too small to check?

The count is reported, the words below threshold appear, and nothing else. No differential, no direction, no ratio. Nineteen of 31 groups fell below a stated threshold of six here, covering 89 people, and all nineteen are named so next cycle's owner does not assume they were cleared.

How is a correction priced?

To the bottom of the comparator range, not to the midpoint. Priya's comparators sit between 98 and 109 percent of midpoint, so the target is 98.0. Midpoint costs roughly three times as much and creates a new outlier somebody explains next April. Levels come from the job description.

Why does every zero award need a written reason?

Because the row is the record, and employment records covering pay rates have to be preserved (29 CFR 1602.14). Budget is not a reason. Nine of 41 zeroes came back for a blank, and four of those nine became small awards once somebody had to write the sentence.

What about someone hired just after the cut-off?

They are out of the cycle and they will ask, so the guidance answers it before allocation opens rather than in the room. If their starting base already looks low, it goes on the correction list rather than being worked around. Starting records come from onboarding.

Find out what your cycle does to band position

Send payroll, the band structure with midpoints, ratings and the budget. The first thing back is every eligible person as a position in band, and the count of job groups where the proposal as submitted widens a differential.

Check my whole proposal