Salary BandingHow to build pay bands that are fair, competitive, and defensible
Most compensation problems are not about budget. They are about structure. Teams overpay new hires to close fast, underpay long-tenured employees who never push back, and make inconsistent offers to people in the same role. Salary bands fix all three, if you build them correctly.
Anatomy of a salary band
Minimum
$80k
Entry to band
Midpoint
$100k
Market target
Maximum
$120k
Band ceiling
New hires
80-95%
Still ramping
Fully performing
95-110%
Solid contributor
Top performers
110-120%
Ceiling territory
Salary banding is not a large-company luxury. Any team hiring more than a handful of people benefits from having a defined pay structure. Without one, you are making compensation decisions on intuition, and intuition compounds inequity over time.
The mechanics are straightforward: group roles into grades, benchmark each grade to the market, then set a minimum, midpoint, and maximum for each grade. What makes bands useful is not the math but the discipline. They force you to define what a role is worth before someone negotiates, not during.
Pay transparency laws in the US are accelerating adoption. States including Colorado, California, Washington, and New York now require salary ranges on job postings. If your bands are not defensible, posting ranges publicly becomes a liability rather than a differentiator.
This guide covers how to build salary bands from scratch, set the right band width, use compa ratios to manage individual positioning, and keep your structure calibrated over time. If you are also working on how offers fit into your broader employer value proposition, start there first so your pay philosophy aligns with your total rewards messaging.
The case for structure
Why most teams need salary bands earlier than they think
The typical sequence is: hire 10 people with no structure, realize you have five people in the same role at very different pay rates, spend two years in awkward retention conversations, then build bands retroactively to clean up the mess. Every team that has gone through this says the same thing: we should have done this at hire 5, not hire 50.
The business case is not complicated. The Society for Human Resource Management consistently reports that pay equity issues are among the leading drivers of voluntary turnover. When employees discover internal pay gaps, the ones who leave first are usually your best performers because they have options. The ones who stay but find out are often quietly disengaged for months before they leave.
Salary bands also solve a hiring speed problem many teams do not recognize. When recruiters do not know the approved range for a role, they go back to hiring managers for approval on every offer. That adds days to your time to hire and creates an approval bottleneck at the worst possible moment. A defined band gives recruiters the authority to move, within guardrails.
For DEI outcomes, the evidence is clear. A Harvard Business Review analysis found that unstructured salary setting systematically disadvantages women and underrepresented minorities, largely because it relies on negotiation behavior that varies by demographic group. Bands do not eliminate bias, but they shift the default from "what did you ask for" to "what does this role pay."
Step by step
How to build salary bands: the six-step process
This is not a one-afternoon project, but it is also not a six-month consulting engagement. A team of 30 to 100 people can build a working pay structure in two to four weeks if someone owns it with actual time blocked.
Six steps from blank sheet to working pay structure
Map and group your roles
Cluster jobs by scope, complexity, and impact into 5-8 grades
Benchmark each grade to market
Pull median pay from 2+ sources (Radford, Levels.fyi, BLS)
Set midpoints at your target percentile
P50 for most roles, P65-P75 for hard-to-fill or technical
Set band width by role type
40-50% for defined roles, 60-80% for senior or specialist
Slot current employees
Calculate compa ratios; identify underpaid, overpaid, and off-grade
Build a correction plan
Phase in adjustments; set a review cadence (annual minimum)
Step 1: Map and group your roles
Start by listing every distinct role in the organization. Not every person, every role. Then group them into grades based on scope of work, decision-making authority, required experience, and impact on the business. A junior software engineer and a senior software engineer belong in different grades. A junior software engineer and a junior data analyst may belong in the same grade if their scope and market value are similar.
Most companies land on five to eight grades across the organization, sometimes more for technical ladders. Fewer grades give you more flexibility but less precision. More grades give you precision but create endless arguments about whether a role is a 4 or a 5.
Step 2: Benchmark each grade to market data
This step is where most teams get stuck because compensation data is expensive or confusing. The options, roughly in order of reliability, are: formal compensation surveys (Radford/Aon, Mercer, Willis Towers Watson), technology-specific data (Levels.fyi), and crowdsourced data (Glassdoor, LinkedIn Salary, Payscale). The Bureau of Labor Statistics Occupational Employment Statistics gives you a free baseline for most job families.
Use at least two sources. Pull the 25th, 50th, and 75th percentile for each role, then anchor your grade midpoints to your target market position. Most growing startups target the 50th to 65th percentile. If you are competing for talent in a hot market, P75 becomes necessary. If you offer strong equity or flexible arrangements that compensate for base pay, P50 can work if you are transparent about the tradeoff.
Step 3: Set midpoints at your target percentile
The midpoint of each band should equal your target market pay for a fully performing employee in that grade. It is not an average of your current employees' salaries. It is what the market pays for that level of work, at the percentile you have chosen to compete at.
If your midpoints are calibrated correctly, a fully effective employee at the midpoint is neither a bargain hire nor overpaid. They are exactly where the market expects someone at their level to be.
Step 4: Set band width by role type
Band width is the range from minimum to maximum, expressed as a percentage of the midpoint. A 50% wide band centered at $100k runs from $75k to $125k. Most organizations use 40-50% for defined, structured roles and 60-80% for senior, technical, or specialist roles where individual contribution varies widely.
Step 5: Slot employees and calculate compa ratios
Once bands exist, place every employee into their grade and calculate where their current salary sits within the band. A compa ratio is simply: (current salary / band midpoint) x 100. An employee at $85k in a grade with a $100k midpoint has a compa ratio of 85. That number tells you everything you need to know about their pay positioning.
Expect to find outliers. You will have people below 80 (likely underpaid relative to market) and people above 120 (may have outgrown their grade, or may have been overpaid at hire). Both require a plan.
Step 6: Build a correction plan
Very few organizations can correct all pay gaps at once. Prioritize by impact: employees who are most below market and in roles where external competition for their skills is highest. Phase corrections into performance review cycles where possible, but do not wait two years to correct a 30% underpayment.
Document your correction criteria. If you decide to bring everyone below 85 compa ratio to 90 within the next review cycle, write that down. Consistency matters for both equity and legal defensibility. For a detailed framework on how to track pay outcomes over time, pair this with your data-driven recruiting metrics.
Structure decision
Narrow bands vs. broad bands: how to choose
This is the design decision that generates the most internal disagreement, and the honest answer is that there is no universally correct answer. It depends on how you manage performance, how much you trust manager judgment, and how much variation exists within each role.
Narrow vs. broad band structures
Narrow bands
40-50% width8-12 grades
Best for: High-volume, clearly defined roles
Easy to explain to employees
Less manager discretion required
Clean for hourly or clerical work
Requires more frequent updates
Limited room for individual variation
Can feel constraining for senior ICs
Broad bands
60-80% width4-6 grades
Best for: Technical, senior, or specialist roles
Accommodates wide experience range
Fewer grade-change conversations
More flexibility for retention
More manager judgment needed
Harder to explain to employees
Equity risk without clear criteria
Broadbanding, where you collapse multiple grades into one wide range, is popular among fast-growing tech companies because it reduces the number of conversations about whether someone is a P3 or a P4. The tradeoff is that it requires strong manager calibration to avoid pay decisions becoming arbitrary. Without clear criteria for what justifies positioning someone at 90% versus 115% within the same broad band, managers fill that void with gut feel, which often replicates existing biases.
My view: narrow bands for your highest-volume roles, broader ranges for IC tracks above senior level. The goal is precision where roles are well-defined and flexibility where individual contribution genuinely varies.
Managing positioning
Compa ratios: the number that actually tells you where your team stands
Compa ratio is short for comparative ratio. It measures where an individual's pay sits relative to the midpoint of their band. The formula: current salary divided by midpoint, multiplied by 100. A 100 compa ratio means the person is paid at market midpoint for their role and level.
Compa ratio distribution check
Role A (Grade 3)
Under$78k actual vs $100k midpoint
Compa ratio
Role B (Grade 3)
On target$97k actual vs $100k midpoint
Compa ratio
Role C (Grade 4)
Above mid$118k actual vs $105k midpoint
Compa ratio
Role D (Grade 4)
Overpaid?$135k actual vs $105k midpoint
Compa ratio
In practice, use compa ratios to answer three questions before your annual review cycle opens:
Who is below 80? Those are likely market correction priorities, especially in roles with active external hiring.
Who is above 120? Those employees have hit the band ceiling. The right response is either a grade promotion (if the role has grown) or a clear conversation about why they are not moving above that number.
What is the average compa ratio by department, gender, or ethnicity? If you see a pattern, you have an equity problem that individual manager intent will not fix.
Most HR teams run compa ratio analysis once per year, right before merit increases. The teams that get ahead of retention problems run it at least twice: once at the start of the year to set priorities, and once mid-year when the job market is active and competitors are making offers.
Tracking compa ratios also feeds directly into quality of hire analysis. If your highest-performing employees consistently drift to above-midpoint positioning while lower performers cluster at the bottom of bands, your merit process is working. If there is no pattern, your merit process is noise.
Staying current
How to keep salary bands from going stale
Bands built from 2022 data are not bands. They are a historical artifact. Pay structures decay faster than most organizations admit. The 2021 to 2023 tech hiring surge moved software engineering pay 20 to 30% in some markets in under two years. Organizations that updated bands annually found themselves 15% below market by mid-year. The ones that updated every two years found their senior engineers getting offers 40% above what the band allowed.
A practical maintenance cadence looks like this:
Annual full review: Pull fresh market data, update midpoints where the delta exceeds 5%, and adjust minimums and maximums accordingly.
Mid-year spot check: Review compa ratios for any grades where you are actively hiring or seeing unusual attrition. If the market has moved more than 8% in a grade, do not wait for the annual cycle.
Trigger-based reviews: Any role where you lose three or more people in six months likely has a compensation problem. Run the benchmark before writing the problem off as a cultural issue.
When you update midpoints, do not automatically adjust current employee pay. Calculate what the update does to compa ratios first. If a midpoint increases by 8%, everyone in that grade loses 8% compa ratio points without any action on your end. That is useful information. It tells you who is now below your target range and needs a correction.
The communication side of band updates is often harder than the data side. Employees who are above the new midpoint should not receive a pay cut, but they also should not expect merit increases until the band catches up to them. Being clear about this proactively is far less damaging to morale than letting people find out indirectly. When paired with a strong offer letter process, bands give you a natural touchpoint to set expectations at hire. See our guide on how to write an offer letter for how to communicate range context to candidates.
What goes wrong
The five most common salary banding mistakes
1. Building bands around current employee pay
The most common error is surveying what you currently pay, averaging it, and calling that a midpoint. That just codifies existing pay inequities into a formal structure. Midpoints should come from market data, not internal history.
2. Too many grades
Organizations that create 12+ grades spend more time arguing about grade placement than doing useful work. Grade placement becomes a proxy status battle. Keep it simple. If two grades cover roles with similar market value, they probably should not be separate grades.
3. Ignoring total compensation
Base pay bands without a clear view of equity, bonus, and benefits create confusion during recruiting. A candidate comparing two offers will look at total compensation. If your base is at P50 but your equity is at P75, that story needs to be told. Build it into your employer value proposition explicitly.
4. No process for overpaid employees
Employees above band maximum are common, usually because they received large raises before bands existed or because a band was updated and their pay did not move. Calling them "red-circled" and doing nothing is not a plan. Have a clear policy: above-band employees are pay-frozen until the market catches up, they move to a higher grade if their role has grown, or there is an explicit exception with documentation.
5. Not communicating the structure to employees
Salary bands built in secret and used only by HR accomplish little. Employees need to understand where they sit in the band, what moves them within it, and what it takes to move to the next grade. You do not have to disclose exact salaries of other employees, but explaining the framework builds the trust that makes pay transparency laws less disruptive.
Frequently Asked Questions
What is salary banding?
Salary banding is the practice of grouping jobs into pay grades, then setting a minimum, midpoint, and maximum for each grade. The band defines the range a person in that role can be paid. It keeps compensation consistent, defensible, and tied to market data rather than individual negotiation outcomes.
How wide should a salary band be?
Most organizations use band widths between 40% and 80% of the midpoint. Narrow bands (40-50%) work well for high-volume, clearly defined roles where the work is fairly uniform. Wider bands (60-80%) make sense for senior or specialist roles where there is meaningful variation in experience and contribution. Broadbanding collapses multiple grades into one wide range, which gives flexibility but requires more manager judgment.
What is a compa ratio and how do I use it?
Compa ratio is an employee's current salary divided by the midpoint of their band, expressed as a percentage. A compa ratio of 100 means they are paid exactly at midpoint. Below 80 usually signals underpayment relative to market; above 120 is red flag territory that often indicates a role outgrown its band. Use compa ratios during annual reviews to identify who is due a market correction before they find out on Glassdoor.
How do salary bands reduce bias in hiring?
Without bands, starting salaries are largely determined by what someone asks for or what a recruiter thinks they can negotiate. That disadvantages people who do not negotiate aggressively, which correlates with gender, race, and background. A defined band anchors offers to the role and level, not the candidate's bargaining skill. The practice alone does not eliminate bias, but it makes the starting point consistent.
How often should salary bands be updated?
Most organizations update bands annually, typically aligned with the compensation planning cycle. In periods of rapid wage inflation, like the 2021 to 2023 tech market, annual updates are not fast enough and you will find your bands trailing market within six months. A good habit is to run a quick benchmark check mid-year and flag any grades where your midpoint has drifted more than 10% below the market median.
What data sources should I use for salary benchmarking?
The most reliable sources are compensation surveys like Radford (Aon), Mercer, and Willis Towers Watson, which require a subscription and participation. For smaller teams, a combination of Levels.fyi (technology roles), LinkedIn Salary, Glassdoor, and BLS Occupational Employment Statistics gives a reasonable baseline. Use at least two sources and weight them by how closely the job definition matches your roles.
Resources
Prepzo guides
Hiring with a cleaner process starts here
Prepzo gives hiring teams the structure to screen, interview, and make offers consistently, without spreadsheet chaos or approval bottlenecks.
Try Prepzo free