River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Salary Band Template with Market Data

Five documents and four sheets, including a register that separates people who are genuinely underpaid from those whose band was priced off the wrong job.

Free download  ·  No account needed

What a Match Grade Is

The column every band template leaves out

You bought a survey. Somebody then decided which survey job each of your job codes corresponds to. That decision became a midpoint, the midpoint became a band, and the band became a number in a conversation. Nothing on the sheet records how good the decision was.

Exact

The survey definition and your work are the same job. Support Engineer priced off Technical Support Engineer, 731 incumbents behind the figure. Name the survey, the cut and the percentile. There is nothing else to explain.

Adjacent

The scope differs in a way that fits in one sentence, and the sentence gets written down now rather than when somebody asks. Enablement Specialist priced off Technical Trainer: the survey job trains employees internally, yours trains customer warehouse staff on site.

Constructed

No survey job exists, so the band was assembled from one that is close, or from a close job plus a premium. This is a judgement expressed to the nearest five hundred dollars. It is usable, and it is not a market rate, and the difference is what the grade is for.

None

Nothing comparable was in the cut. The band is internal relativity, which is a defensible thing to be and an indefensible thing to hide.

The grade is not documentation. It decides what a manager may claim, which bands get re-checked before the next merit cycle, and above all how to read somebody sitting below their floor: on an Exact band that is a pay decision, and on a Constructed band it is a mapping error until proven otherwise.

A band review ends with a list of people below the floor and a bill to lift them. At Marloe Logistics, a fictional 240-person warehouse software company, that list ran to 19 people and the bill came to 153,600 dollars. Eight of the 19 were being measured against bands assembled out of survey jobs that describe different work. Re-checking those against job content put five of them inside band with nothing spent, and the honest figure was 88,600.

So 65,000 dollars of the first number, 42.3 percent of it, was a mapping error queued for funding as though it were underpayment. It buys nothing anybody can point at, and it tells five people their pay was wrong when the band was. Whether two jobs are comparable turns on actual job requirements rather than titles, so a code with no written job description is a code that gets matched on its title and priced on that.

The correction runs both ways, which is the part worth trusting. One code had been priced off a survey job with no engineering scope at all, so its band moved up. Ronan Kelleher's gap went from 4,500 to 15,500, the second smallest on the original list and the largest on the finished one. A mechanism that only ever finds savings is a budgeting trick with a methodology attached. This one hands a merit cycle a structure rather than a saving.

The same 19 people, priced at 153,600 and then at 88,600

The Market Comparison, the Band Structure behind it, and the Outlier Register priced both ways.

Market Comparison

Illustrative for a fictional warehouse software company, Marloe Logistics. FY27, 31 job codes, survey effective 1 April 2026 aged forward 2.8 percent to bands effective 1 January 2027.

CodeInternal titleSurvey job it was priced fromGradenAged
50th
What can be said to an employee
SFT-ENGSoftware EngineerSoftware Engineer, ApplicationsExact1904124,000The survey, the cut and the percentile
ACC-MGRAccount ManagerAccount Manager, SoftwareExact843112,000All of that, and that the band is base only
SUP-ENGSupport EngineerTechnical Support EngineerExact73179,000The survey, the cut and the percentile
IMP-CONImplementation ConsultantImplementation ConsultantExact412108,000The survey, the cut and the percentile
SOL-ARCSolutions ArchitectPre-Sales Solutions ConsultantAdjacent289126,000All of that, plus: survey job is pre-sales only, ours owns post-sale design
ENB-SPCEnablement SpecialistTechnical TrainerAdjacent17486,000All of that, plus: survey job trains internally, ours trains customers on site
WMS-IMPWMS Implementation LeadImplementation Consultant plus a lead premiumConstructedn/an/aThat it was constructed, from what, and that it is under review
EDI-SPCEDI Integration SpecialistIntegration EngineerConstructedn/an/aSame, and their own figure counts as evidence in that review
LOG-DATLogistics Data EngineerData EngineerConstructedn/an/aSame
CST-ENGCustomer Success EngineerTechnical Account ManagerConstructedn/an/aSame
FLD-SVCField Service CoordinatornoneNonen/an/aThat there is no market figure, and what the internal comparison was
31 codes: Exact 14, Adjacent 9, Constructed 6, None 2. Eleven shown. The distribution is reported before any salary figure appears, because it is the first true thing anybody learns about their band structure.

The n column is not decoration. A band priced off 174 incumbents and a band priced off 1,904 carry different weight, and whoever quotes the band in a room next quarter needs to know which one they are holding.

Band Structure by Level and Location

Sample rows from 76. Every band keeps the midpoint as first built beside the midpoint it would carry if the match were re-graded, because collapsing those two columns is what makes a mapping error look like underpayment.

CodeLvlLocationFactorMinimumMidpointMaximumGradeMidpoint
re-graded
IMP-CONL2Columbus OH0.9385,500100,500115,500Exactno change
IMP-CONL2Reno NV0.9789,000104,500120,000Exactno change
IMP-CONL2Remote, national1.0092,000108,000124,000Exactno change
SUP-ENGL2Columbus OH0.9362,50073,50084,500Exactno change
WMS-IMPL2Columbus OH0.93103,000121,000139,000Constructed106,000
WMS-IMPL3Reno NV0.97132,000155,000178,000Constructed136,000
EDI-SPCL2Columbus OH0.9392,000108,000124,000Constructed95,000
LOG-DATL2Reno NV0.97114,000134,000154,000Constructed118,500
CST-ENGL2Columbus OH0.9382,00096,500111,000Constructed109,500
CST-ENGL3Reno NV0.97103,500122,000140,500Constructed138,500
Range width is 85 to 115 percent of midpoint on every band in the company, so a position in band means the same thing in Support as it does in Engineering.

Three constructed codes come down and one goes up. The location factor is the survey's own cost-of-labour cut, which spans seven points across these three markets. It is not a cost-of-living index: that measures what it costs to live somewhere, which is a different quantity with a wider spread, and it is not what these bands are built on.

Outlier Register

Nineteen below the floor and seven above the ceiling, out of 240. Every row priced twice, and no base pay reduced anywhere.

EmployeeCodeGradeBaseFloor as
first built
Cost to
that floor
Floor once
graded
Cost to the
graded floor
What to do
Aisha DanjumaSUP-ENGExact58,30062,5004,200no change4,200Fund it. The match holds.
Tobias RehnDAT-ANLExact67,40074,5007,100no change7,100Fund it. The match holds.
Gordon AchebeSFT-ENGExact121,000129,0008,000no change8,000Fund it. The match holds.
Noor RahimiSOL-ARCAdjacent125,500132,0006,500no change6,500Fund it. The match holds.
Declan MoriartyWMS-IMPConstructed94,500103,0008,50090,0000Nothing. The band was wrong, the pay was not.
Sipho NdlovuWMS-IMPConstructed118,000132,00014,000115,5000Nothing. The band was wrong, the pay was not.
Ana Beatriz ReisLOG-DATConstructed128,000143,00015,000126,0000Nothing. The band was wrong, the pay was not.
Malik OseiLOG-DATConstructed98,000114,00016,000100,5002,500Fund it, at the corrected figure.
Ronan KelleherCST-ENGConstructed77,50082,0004,50093,00015,500Fund it, at the higher corrected figure. The band was too low.
Total below band19153,60088,60065,000 of the first figure was a mapping error, 42.3 percent of it.
Wendell GrovesSUP-ENGExact112,000106,0000no change0Held at 112,000. Nineteen years in the job.
Priyanka DeviCST-ENGConstructed145,000140,5000159,5000Inside the corrected band. No conversation needed.
Nkechi BalogunSFT-ENGExact182,000175,0000no change0Inside the L4 band. To levelling, not to payroll.
Total above band7002 inside the corrected band, 3 held on tenure, 2 to levelling.

Ronan is the row that makes the rest of the register believable. Correcting the four constructed codes cleared five below-band rows at no cost and closed two above-band ones, and it also tripled the largest genuine gap in the company. The nine rows on Exact and Adjacent bands never moved: their matches hold, so their gaps were always real.

What is in the pack

01

Market Comparison

One row per job code: the survey job it was priced from, the match grade, the scope difference in a sentence, how many incumbents sit behind the figure, and what may be said to an employee.

02

Band Structure by Level and Location

Minimum, midpoint and maximum for every job code, level and location, each with its match grade and the midpoint it would carry if that match were re-graded, side by side.

03

Current Position in Band

Everybody as a percentage of their own band midpoint rather than as a salary, carrying the grade of the band they are measured against. Three statuses and no traffic lights.

04

Outlier Register

The out-of-band rows priced twice, against the bands as first built and against the graded ones, in decision order: what needs funding, what needs re-checking, what needs levelling.

05

What a Match Grade Is

The two pages to read before the Band Structure sheet, on the four grades and on why the out-of-band list means nothing until it is split by them.

06

Compensation Philosophy

A page of decisions rather than values, every sentence one somebody could disagree with. What you pay against, what a band is, and what happens to people outside one, including the starting bases that created a good share of them. Later, the total rewards statement reports each midpoint back to the person sitting on it, and a payroll audit checks that the rate actually paid still matches what this page approved.

07

Band Rationale

One entry per job code, ending with the sentence somebody needs at short notice in a room: what may actually be said about this band, which depends entirely on its grade.

08

Manager Guidance and Answering a Published Salary Number

What each grade lets a manager claim, the four conversations that happen, and the four-step reconciliation for somebody who found a bigger figure than their salary.

How it works

  1. 1

    Send the survey and your job codes

    The survey extract with its job definitions, your codes with levels and locations, and current base pay per person. Job descriptions help most for the roles you invented.

  2. 2

    Every match gets graded

    Each code is matched against survey job definitions rather than survey job titles, then graded Exact, Adjacent, Constructed or None. The distribution is reported before any salary figure appears.

  3. 3

    Bands get built, everybody gets placed

    Midpoints aged forward on a stated assumption and cut by location, one range width company-wide, then everybody expressed as a position in their own band.

  4. 4

    The outlier list gets split before it gets priced

    Rows on graded matches are pay decisions. Rows on constructed bands are re-checked against job content first, and priced afterwards at whichever figure the correction produces.

Frequently asked questions

Why does the match grade change what a band costs?

Because a person below a band built on a constructed match is a mapping error before they are a pay problem. Eight of Marloe's 19 below-band people sat on constructed bands. Re-checking those four codes moved five inside band with nothing spent, taking the bill from 153,600 to 88,600.

Is this just a way to make the remediation number smaller?

One of the four constructed codes had been priced off a survey job with no engineering scope, so its band went up. Ronan Kelleher's gap went from 4,500 to 15,500, the largest in the company. The nine rows on Exact and Adjacent bands never moved at all.

What do I tell someone who found a bigger number on a salary site?

Four steps, on paper in front of them. One Marloe employee found 132,000 against her 104,000 base. Bonus accounted for 12,000, geography 8,400, company size 6,700, and the real difference was 904 dollars. The pack works that case all the way through.

Why cut bands by location at all?

Because the comparison that matters legally is ordinarily within a distinct physical place of business, so location is structural rather than an adjustment applied afterwards. The factor is the survey's own cost-of-labour cut, seven points across three markets, not a cost-of-living index.

What happens to people paid above the band maximum?

No base pay is reduced. Two of Marloe's seven were only above because the band was too low. Three are held at their rate on long tenure, which is a differential maintained for reasons unrelated to sex. Two are above because their level is wrong, so they go to levelling.

How is this different from a pay equity analysis?

This builds the structure and reports who sits outside it. A pay equity analysis looks backwards at differences between comparable people and separates the explained from the unexplained. Run on a band structure whose matches are ungraded, it produces findings nobody can act on.

What format are the downloaded files, and is it free?

Five Word documents and four CSV sheets in one zip, free and with no account. The documents open in Word, Pages and Google Docs, the sheets in Excel, Numbers and Google Sheets. Edit with AI installs the same pack as a private Space and grades your own matches.

Find out which of your gaps are mapping errors

Send the survey extract with its job definitions, your job codes with levels and locations, and current base pay. The first thing back is the grade distribution across your codes, before any salary figure, then the out-of-band list priced both ways.

Edit with AI