Salary Bands: What They Are and How to Build Them
What a salary band is, the min-mid-max formula, how to build bands from free market data, and how to fix an employee who is already outside the range.
Salary Bands
What they are, how to build them from free data, and how to fix the people already outside them
A salary band is a defined pay range for a role or level, made of a minimum, a midpoint, and a maximum. It is the answer to the question "what do we pay for this job," written down in advance, before a specific person is sitting across the table asking for a number.
That last clause is the whole point, and it is why I started using them. Without bands, every salary is a negotiation, and negotiations are not won by the best employee. They are won by the best negotiator. I have paid two people doing the same job meaningfully different amounts, and the only honest explanation was that one of them asked and the other did not.
This guide covers what a salary band actually is, the formulas nobody shows you, how to build defensible bands from free data when you cannot afford a $30,000 survey subscription, a full worked example for a 20-person company, and what to do about the person who is already paid outside the band you just built.
What Is a Salary Band?
The words band, pay band, and compensation band all mean the same thing. There is no meaningful distinction, and anyone drawing one is selling you something.
Anatomy of a Band
The critical thing to internalize: you set the midpoint first. It is the market rate, and it comes from data. The minimum and maximum are derived from it. Most people who build bands badly do it the other way around, picking a low number and a high number that feel right, which produces bands with wildly inconsistent widths and no defensible logic behind any of them.
Band vs Grade vs Range vs Scale
These four terms get used loosely, and the loose usage is mostly harmless, but it is worth knowing what each properly means.
| Term | What It Actually Means | Example |
|---|---|---|
| Pay grade | The label for a level. It is a name, not a number. | Level 3, Grade 7, Senior II |
| Salary band / pay band | The dollar range attached to a grade or level. Min, midpoint, max. | $62,500 to $87,500, midpoint $75,000 |
| Pay range | Often used to mean the range for one specific job, rather than a level covering several roles. | The range for 'Operations Manager' specifically |
| Pay scale / salary structure | The whole system: every level and its band, together. | Your complete set of four levels and four bands |
| Broadbanding | Consolidating many narrow bands into a few very wide ones. | Replacing eight levels with three wide bands |
In practice you define your levels first, then attach a band to each. Level is the container, band is what is in it.
Why a Small Business Actually Needs Them
The standard answer is fairness and compliance. Both are true, and neither is what makes a founder actually build them. Four reasons that are.
| Problem Without Bands | What Bands Do About It |
|---|---|
| Pay is decided by who negotiates hardest | The band decides the range before anyone negotiates. The candidate's leverage is capped by a number you set in advance, in a calm moment. |
| You cannot answer 'why does she earn more than me' | You can. She is further through the band, for reasons tied to scope and performance that you wrote down. This conversation is unwinnable without a structure. |
| Your job postings are non-compliant | Pay transparency laws increasingly require a range in the posting. A band is the range. Without one, you are inventing a number under pressure, which is exactly how you end up with a $50,000 to $200,000 posting that satisfies nobody. |
| Small pay gaps compound into indefensible ones | Merit increases are percentages, so an unexplained gap in year one is a larger unexplained gap in year five. Bands catch it while it is still small. |
The third one is increasingly the forcing function. The fourth is the one that quietly becomes a legal problem: under the Equal Pay Act, a pay difference between employees doing substantially equal work must be explained by a legitimate factor, and the burden of showing that sits with you. A documented band structure is one of the cleanest ways to meet it. The pay equity guide covers what else makes a pay difference defensible.
The Formulas
Most articles on this topic describe bands without ever showing you how to compute one. Here is the arithmetic.
Two notes. The min formula looks odd until you realize what it is doing: it positions the midpoint correctly between the min and max, rather than at the arithmetic average of them, which is not the same thing when the band is expressed as a percentage spread off the minimum.
And compa-ratio is the number to actually manage with day to day. It normalizes across every role in the company: a 0.85 means the same thing whether the band is $50,000 or $250,000. The compa-ratio guide covers how to use it in a review cycle.
Getting Market Data Without a Survey Subscription
Every article on salary bands assumes you have Mercer or Radford data. At a 20-person company you do not, and you are not going to, and the honest answer is that you do not need it to build a first set of defensible bands.
The BLS Occupational Employment and Wage Statistics program is the workhorse here. It gives you median and percentile wages by occupation and by metropolitan area, it is free, and it is federal government data, which means when someone asks how you set the number, you have an answer that does not sound like a guess.
One caveat on what you are actually comparing. A salary band is a wage range, not a total compensation figure. Per the BLS Employer Costs for Employee Compensation data, benefits account for roughly 30 percent of total compensation for private industry workers, so the loaded cost of a person sitting at your $75,000 midpoint is meaningfully higher than $75,000. Build the band on wages, but budget on the loaded figure. The labor cost guide covers the full build.
How to Build Your First Bands
A Worked Example
Every number in that table comes out of the formula. Nothing is invented. If you disagree with a midpoint, that is a data argument you can have, which is exactly the property you want: the structure is arguable on its merits rather than on who feels strongly.
Choosing Spread and Midpoint Progression
Two design decisions define the shape of your whole structure.
Spread: How Wide Each Band Is
| Level | Typical Spread | Why |
|---|---|---|
| Entry level | 25% to 35% | Less room for individual variation. An entry-level person is either doing the job or learning it, and the market price band is narrow. |
| Mid level / professional | 35% to 45% | Real variation in capability and market price emerges. The band has to hold both a competent person and a strong one. |
| Senior / lead | 45% to 55% | The gap between an adequate and an outstanding senior person is enormous in both output and market price. A narrow band cannot hold both. |
| Executive | 50% to 60%+ | Widest, and increasingly individual. At this level the band is more of a guardrail than a grid. |
Midpoint Progression: The Gap Between Levels
Midpoint progression is how much each level's midpoint exceeds the one below it. Common practice is 10 to 20 percent.
At 15 percent progression starting from a $55,000 midpoint, your four levels land at roughly $55,000, $63,250, $72,750, and $83,650. Too small a gap and a promotion is not worth taking. Too large and you cannot afford to promote anyone, so you do not, and they leave.
When Someone Is Already Outside the Band
You will build your first bands and immediately discover that two or three people do not fit in them. This is not a failure of the exercise. It is the exercise working.
Below the Minimum (Green-Circled)
Someone paid below the band minimum is underpaid relative to your own stated market position. This is a market adjustment, not a merit increase, and confusing the two is a mistake I have watched a lot of founders make.
A 3 percent merit increase does not close a 15 percent gap. It closes a fifth of it, it makes the person feel rewarded, and it leaves them underpaid, which makes the eventual discovery worse rather than better. Fix the base first, with a market adjustment, and run merit on top of the corrected number. The merit increase guide covers why these are different instruments.
Above the Maximum (Red-Circled)
Someone paid above the band maximum has three possible resolutions.
| Option | What You Do | When It Is Right |
|---|---|---|
| Freeze and let the band catch up | Hold their base pay while annual band adjustments raise the range toward them. Give any increase as a one-time bonus rather than base pay. | The person is genuinely at the top of this role's value and there is no bigger role for them. Resolves in two to four years. |
| Promote them into the next band | Their scope has actually outgrown the role and the band is telling you so. | Very common. The red circle is frequently the structure noticing a promotion that was overdue. |
| Fix the band | Rebuild the midpoint from data. If the person is well-regarded and the market says they are paid correctly, your band is wrong. | Check this before you conclude the person is overpaid. A single red-circled person might be an outlier. Three of them means your data was bad. |
Bands and Pay Transparency Law
Pay transparency law is the reason a lot of small businesses are reading about salary bands for the first time. You cannot post a good-faith salary range if you have never decided what your range is.
Two practical consequences worth internalizing.
Remote postings can trigger multiple states. Several states apply their posting rules to remote roles that could be performed in the state. One remote job posting can therefore trigger several states' requirements at once, and the safe approach for a multi-state or remote-first employer is to comply with the strictest applicable rule.
A good-faith range means a real range. Posting $50,000 to $200,000 to technically satisfy a law is not a good-faith estimate, and several states say so explicitly. A band gives you a range you can actually defend, because it was derived from data before anyone asked. The pay transparency laws guide covers the state-by-state picture.
Broadbanding and Government Pay Bands
Two adjacent concepts worth briefly disambiguating, because they are what people find when they search this term and get confused.
Broadbanding is the practice of collapsing many narrow bands into a few very wide ones. Instead of eight levels each with a 30 percent spread, you might have three bands each with a 100 percent spread. It buys flexibility and it costs structure, and it tends to work in organizations with strong management judgment and to fail in organizations without it, because a very wide band means the manager, not the structure, is deciding pay. For a small business it is usually the wrong starting point: you want more structure at the beginning, not less.
Government pay bands are a different animal. The US federal General Schedule uses grades and steps rather than the min-mid-max bands described here, and some federal agencies have adopted broader pay banding systems that consolidate GS grades. If you landed here looking for how the GS scale works, that is a separate system, and none of the formulas above apply to it.
Should You Share Bands With Employees?
The honest answer is that you are going to, whether you decide to or not.
If you post a range in a job ad, your current employees will see it, and they will compare it to their own salary. If your bands are not published internally, they will be reconstructed from that job posting, in the least favorable possible interpretation, in a conversation you are not part of.
Note also that under the National Labor Relations Act, employees have a protected right to discuss their pay with one another. A policy prohibiting it is unlawful, and attempting to enforce one is a bad idea for reasons that go beyond the legal exposure. The Department of Labor maintains an overview of the equal pay protections that sit behind this.
So the real choice is not whether your pay structure becomes known. It is whether it becomes known as something you built deliberately and can explain, or as something inferred from a job ad. I would rather be explaining a structure than defending an inference. The workplace transparency guide covers how to stage this without doing it all at once.
Where this gets operationally messy at a small company is that the band lives in a spreadsheet, the employee's actual salary lives in payroll, the promotion that moved them between bands lives in someone's email, and nobody can produce a compa-ratio without an hour of assembly. That is the gap FirstHR closes: employee profiles that hold the band, the level, and the compensation history together, so the answer to "where is this person in their range" takes ten seconds. It does not run your payroll or set your bands. It holds them where the rest of the employee record lives. The HRIS guide covers what else belongs in that system.
Common Salary Band Mistakes
| Mistake | What Happens | The Fix |
|---|---|---|
| Building the band before setting the midpoint | You pick a low and a high number that feel right, and end up with bands of wildly inconsistent width and no defensible logic. | Set the midpoint from market data first. Derive min and max from a chosen spread. The midpoint is the anchor. |
| Creating a level for every person | You have not built a structure, you have documented the status quo, and you cannot promote anyone into anything. | Three or four levels at 20 people. Group by scope and impact, not by title or tenure. |
| Using a merit increase to fix a below-minimum salary | A 3% merit increase does not close a 15% gap. The person stays underpaid and now also feels rewarded. | Market adjustment first, to bring them to the band minimum. Then run merit on the corrected base. Different instruments. |
| Assuming you need a paid compensation survey | You conclude you cannot afford bands and do nothing, which is strictly worse than imperfect bands. | BLS wage data plus 10-15 live job postings in your market. An afternoon, no cost, and defensible. |
| Cutting pay to force someone into a band | Legally fraught and a guaranteed resignation. You have solved a spreadsheet problem by creating a hiring problem. | Freeze and let the band catch up, promote, or fix the band. There is no fourth option. |
| Same spread at every level | Your senior band is too narrow to hold both an adequate and an outstanding senior person, so you break it constantly. | Widen with seniority. Roughly 30% entry, 40% mid, 50% senior. |
| Setting midpoint progression too small | A promotion nets less than a good merit increase, which makes promotion a demotion in cash terms. | 10% to 20% progression between levels. Test it: is the promotion raise clearly bigger than the merit raise? |
| Never updating the bands | The market moves, your bands do not, and you start losing candidates at offer stage without understanding why. | Rebuild midpoints from fresh data annually. Check who has fallen below the new minimum. |
Frequently Asked Questions
What is a salary band?
A salary band is a defined pay range assigned to a role or job level, consisting of a minimum, a midpoint, and a maximum. It sets the floor and ceiling of what your company will pay for that role. The midpoint represents the market rate for someone fully competent in the job, the minimum is where a new person starts, and the maximum is the ceiling for someone excelling in the role. Salary bands exist so that pay decisions are consistent, defensible, and aligned to the market rather than to whoever negotiates hardest.
What is a pay band?
A pay band is the same thing as a salary band: a defined range with a minimum, midpoint, and maximum assigned to a role or level. The terms are used interchangeably, along with compensation band. You will also encounter the term in a government context, where the US federal General Schedule uses pay grades and steps, and some agencies use broader pay banding systems that consolidate several grades into a single wide band.
What is the difference between a pay band and a pay grade?
A pay grade is the label for a level, such as Level 3 or Grade 7. A pay band is the actual dollar range attached to that level. In practice a company defines its levels first, then attaches a band to each. A pay range usually refers to the range for one specific job, while a salary band is broader and can cover several similar roles at the same level. A pay scale is the whole structure of all your bands together.
How do you calculate a salary band?
Start with the midpoint, which you set from market data, then derive the minimum and maximum from a chosen spread. The formula is: minimum equals midpoint divided by (1 plus spread divided by 2), and maximum equals minimum multiplied by (1 plus spread). For a $75,000 midpoint at a 40 percent spread, the minimum is $62,500 and the maximum is $87,500. Do not build a band by taking a number and adding and subtracting an arbitrary amount, because the resulting spread will be inconsistent across roles.
What is a typical salary band spread?
Spread is the width of the band, calculated as (maximum minus minimum) divided by minimum. It typically runs 25 to 35 percent for entry-level roles, 35 to 45 percent for mid-level and professional roles, and 45 to 60 percent for senior and executive roles. The spread widens with seniority because the performance and market-price gap between an adequate and an outstanding senior person is much larger than at entry level. Midpoint progression between adjacent levels is commonly 10 to 20 percent.
How do you build salary bands without a compensation survey?
Triangulate free sources. Start with the BLS Occupational Employment and Wage Statistics, which gives median and percentile wages by occupation and metro area, and is government data you can defend. Cross-check against 10 to 15 live job postings for the same role in your market, which is easier now that many states require a posted range. Sanity-check against a self-reported source such as Glassdoor or Levels.fyi. Where the sources converge is your midpoint. This takes an afternoon and costs nothing.
What is compa-ratio?
Compa-ratio is an employee's actual salary divided by the midpoint of their band. A compa-ratio of 1.0 means they are paid exactly the market rate for a fully competent person in that role. Below 0.90 usually signals someone new to the role or someone who is underpaid. Above 1.10 signals a strong performer approaching the top of their band. It is the single most useful number for spotting pay problems, because it normalizes across roles: a 0.85 is a 0.85 whether the band is $50,000 or $200,000.
What do you do if an employee is paid above the band maximum?
This is called being red-circled, and you have three options. First, freeze their base pay and let annual band adjustments gradually bring the band up to them, which resolves it over a few years. Second, promote them into a higher band if their actual scope justifies it. Third, verify your band is correct, because sometimes the red circle is telling you the band is too low rather than the person too expensive. What you should not do is cut their pay, which is legally risky and practically guarantees you lose them.
Do I have to share salary bands with employees?
No federal law requires you to publish internal bands to your staff, but pay transparency law is moving fast at the state level and increasingly requires a range in job postings, which means candidates and your own employees will see it anyway. Note also that the National Labor Relations Act protects employees' right to discuss their pay with each other, so a policy forbidding it is unlawful. The practical position is that your bands will become known. It is better to publish them deliberately than to have them reconstructed from a job posting.
How often should salary bands be updated?
Review them annually at minimum. Market rates move, and a band that was accurate two years ago quietly becomes a recruiting problem: you keep losing candidates at the offer stage and cannot work out why. The usual cause is that the market moved and your band did not. Rebuild the midpoints from fresh data once a year, and check whether anyone has drifted below the new minimum, because that group needs an adjustment rather than a merit increase.