Can market data set your salary bands?
Benchmark data tells you what other employers pay. Where your bands sit is a choice the data can inform but not make.
Short answer: Not on its own. Benchmark data shows what a chosen set of employers pays for jobs matched to yours. Building salary bands from benchmark data still takes choices the data cannot make: Which market you compete in, where you want to sit against it, how good each job match is and how the bands fit your levels. Those choices move the midpoint more than the data does.
This is general information, not legal advice. Laws and guidance are summarized as checked on 20 October 2026; each one links to its source below.
The bands nobody chose
Northwind Analytics is a fictional 120-person data company based in Denver that hires remotely across the US. Last year its finance team bought a subscription to a compensation benchmarking tool. The HR lead wanted salary bands before the company started posting pay ranges, so she did what the tool made easy. She matched each role to the closest job in the tool, took the market median for each one and used it as the midpoint of the band.
The bands were done in a week. Within a month, two people were arguing about them.
The head of engineering said the bands for senior data engineers were too low to hire anyone. His last two finalists had offers from much larger companies, well above the band. The finance lead said the opposite about customer support. Most of the support team was paid below the new minimums. Fixing that would cost more than the year's raise budget for the whole department.
Both of them pointed at the same data. Neither was wrong about it. The tool's default peer group was US companies of 100 to 500 employees. That was a fair picture of who Northwind competed with for support staff, who were hired locally. It was the wrong picture for senior engineers, who were being hired away by companies ten times Northwind's size. And several support roles had been matched to jobs one level above what the team actually did.
Nobody at Northwind had decided to pay the median of small US companies for every job. The tool's settings had decided it for them. The data was accurate. The bands rested on choices nobody had made on purpose.
Northwind Analytics is fictional and its details are illustrative.
Why the data looks like the answer
A benchmarking tool or salary survey returns clean numbers: A 25th, 50th and 75th percentile for each job, often by location. The midpoint of a band is the pay rate the company considers right for a fully competent person in the job. The market median is the pay rate in the middle of the employers in the data set. It is natural to treat the second as the first. That shortcut is where most band problems start.
Two good goals pull against each other. The first is defensibility: A band the company can trace back to outside data, so managers, employees and finance accept it as fair rather than invented. The second is fit: A band that matches who the company really competes with for each job and what it can afford to pay. Copying the data straight into the band feels like the most defensible option. It often gives the worst fit, because the data reflects someone else's choices about which employers count as the market.
Even the idea of a standard target is less solid than it looks. A compensation advisory firm compared two large 2026 surveys in April 2026. In one, a survey of mostly smaller and mid-sized organizations, 45% of respondents targeted the 50th percentile and 17% targeted above it. In the other, which leans toward larger employers, about 84% targeted the 50th percentile for non-executive roles. The firm's explanation was not that one survey was wrong. It was that the answer depends on how each employer defines its market. It also noted that targeting the median of a broad market can produce higher pay than targeting the 75th percentile of a narrow one.
There is a timing problem too. Survey data describes pay on a date that is usually months before the band goes live. WorldatWork reported in July 2026 that US salary increase budgets averaged 3.6% in 2026, with 3.6% projected for 2027, across 1,799 organizations. A year-old figure used as this year's midpoint is already behind the market it claims to describe.
None of this makes the data less useful. It means the data is an input to a set of decisions. Those decisions are the band.
Five ways teams let the data choose the band
Each of these starts from good data. Each one ends with a band the company did not really choose and cannot fully explain.
"The median is the neutral choice"
What it looks like: Every job family gets a midpoint at the market median, because that is the tool's default and nobody wants to argue for anything else.
Why smart teams fall into it: The median feels like no position at all. Paying above it looks generous and needs a budget case. Paying below it looks stingy and needs a different case. The median seems to need no case.
What it costs: The median is a position. Used everywhere, it is the same position for every job. A company that competes hard for a few scarce skills and easily for everything else ends up overpaying in the easy families and losing hires in the hard ones. Northwind's engineering and support bands failed in opposite directions for exactly this reason. The surveys above show how differently employers answer this question. A median midpoint is defensible only when the company has actually decided that the median of that market is where it wants to be for that family.
"The peer group is whatever the tool shows"
What it looks like: The data cut is left at the tool's default: One industry, one size band and national or metro geography for every job.
Why smart teams fall into it: Changing the cut feels like gaming the numbers. Leaving the default in place feels objective. Most tools also make the default the easiest view to export.
What it costs: The peer group decides the number more than the percentile does. Different jobs compete in different markets. Support and operations roles are often hired from the local market and from companies of a similar size. Senior technical, data and leadership roles often compete nationally and against far larger employers. One cut for every job underprices the second group and can overprice the first. The fix is not a more generous cut. It is a deliberate one, chosen per family, that names the employers the company actually loses people to and hires people from.
"A close-enough match will do"
What it looks like: Roles are matched to the tool's jobs by title. A "Senior Analyst" is matched to the tool's senior analyst, whatever the role actually involves. Blended roles are matched to whichever job has the higher number.
Why smart teams fall into it: Matching dozens of roles is slow and the titles look the same. Reading each job description in the survey feels like more care than the task deserves.
What it costs: A match one level off moves the midpoint more than most percentile debates. Titles mean different things at different companies, so a title match can land a role a full level above or below its real scope. Several of Northwind's support roles were matched one level too high, which is part of why so many of the team fell below the new minimums. Weak matches are not wrong to use. They are wrong to use without saying so. A band built on a weak match should carry less weight in decisions until the match is better understood. This is also why bands depend on levels being settled first. Our guide on grades, bands and levels explains how the three connect.
"Last year's data is this year's market"
What it looks like: Bands are built from the latest data available, whatever its effective date. They are then used unchanged until someone notices offers no longer close.
Why smart teams fall into it: The data arrives once a year and the band project happens when it happens. Adjusting for age feels like guesswork layered on data.
What it costs: Bands quietly fall behind the market. With pay budgets running around 3.6% a year, data that is a year old at launch and another year old by the next review describes a market that has moved on. The first sign is usually recruiting: Offers at the top of the band, then above it. Each exception then becomes a reason to distrust the bands. A sound band says what date its data describes and when it will be looked at again.
"We can fill the gaps by asking peers"
What it looks like: The tool has no good match for a few roles, so someone calls contacts at similar companies, joins an informal salary swap or shares the company's own figures in return for theirs.
Why smart teams fall into it: It is quick, the contacts are friendly and the numbers feel more current than any survey. Many HR networks have traded pay information informally for years.
What it costs: This is the trap with legal risk attached. The US Department of Justice and the Federal Trade Commission published antitrust guidelines in January 2025 that address it directly. As published, they say that exchanging competitively sensitive information, including information about compensation and benefits, with companies that compete for the same workers may be unlawful. They add that the risk does not go away when the exchange runs through a third party or an algorithm. The guidelines are agency guidance, not law. Enforcement priorities can also change. The safer reading is still to rely on properly run surveys and tools for market data and to avoid direct swaps with competitors.
What a defensible band rests on
A good band is not one that matches the data exactly. It is one where every gap between the data and the band is a choice someone made and can explain. In our view, sound bands share these qualities.
- Each family has a stated market. The company can name, in a sentence, which employers it compares itself with for that family and why.
- Each family has a stated position. The target against that market is written down, even where it is the median. Different families can have different targets.
- Match quality is known. Strong, partial and weak matches are recorded. Bands resting on weak matches are treated with more caution until the match improves.
- The data has a date. Everyone using the bands knows when the data was effective and when the bands will next be reviewed.
- The bands fit the levels. The steps between levels make sense as a progression. A band that jumps or overlaps strangely is a sign the match or the level, not the data, needs a second look.
- Current pay was tested against the bands. Where many people in one job sit outside the band, the company checks the match and the level before it changes anyone's pay or the band.
- A manager can explain the band. When an employee asks why their range is what it is, the answer is the market and the position the company chose, not "the tool says so".
The last point matters more each year. Massachusetts lets a current employee ask for the pay range of the job they hold. California gives employees the pay scale for their own position on request. Once ranges are posted, staff compare them with their own pay within days. Our guide on readiness for pay transparency covers what has to be in place before that happens.
The calls only you can make
Data and method can narrow these choices. They cannot make them.
Where to sit against the market, family by family. Paying above the median for the few roles that decide the company's success and at the median elsewhere is a common and defensible stance. So is paying below the median in cash and making up the gap with equity, if the equity is real and explained. Either way, it is a statement about what the company values.
Which market counts for each family. Choosing a broader or more senior peer group for a family raises its bands and its cost. Choosing a narrower one keeps costs down and may cost hires. The right answer depends on where the company really recruits and loses people, which leaders know better than any tool.
What to do when the data says people are underpaid. New bands often show some current staff below the minimum. Whether to close that gap in one cycle, over two or three, or by revisiting the match is a budget and fairness call. How it is handled tells employees more about the company than the bands do.
How often to move the bands. Moving bands every year keeps them current and raises expectations that pay will follow. Moving them less often is cheaper and lets them drift. Neither is wrong if the company says which it does.
The data, the law and the numbers
As of 20 October 2026, these are the points we checked that bear on building bands from market data.
Market positioning is not standard. A compensation advisory firm reported in April 2026 that 45% of respondents to one 2026 survey targeted the 50th percentile and 17% targeted above it, while about 84% of respondents to another survey targeted the 50th percentile for non-executive roles. It attributed the gap mainly to how each survey's employers define their market [1].
Pay budgets keep moving the market. WorldatWork's 2026 to 2027 salary budget survey, reported in July 2026, found US salary increase budgets of 3.6% in 2026 and projected 3.6% for 2027, across 1,799 organizations. It had reported 3.7% for 2025 [2]. Its 2025 survey release had reported 3.9% for 2024 [3].
Sharing pay data with competitors carries antitrust risk. The DOJ and FTC guidelines, as published in January 2025, say exchanging competitively sensitive compensation information with companies that compete for workers may be unlawful, including through a third party or an algorithm [4].
Pay transparency laws assume the range exists before the ad. Washington's guidance says the posted range is the lowest to the highest pay set before posting and that an employer without one must create one [5]. California defines the pay scale as the salary or hourly wage range the employer reasonably expects to pay for the position [6]. Massachusetts lets current employees request the range for the job they hold [7]. Illinois guidance warns that an extremely broad range may raise questions under its equal pay law [8]. A band built on data nobody can explain is hard to defend under any of them.
Our view
We think market data should be the most important input to a salary band and never the decision. The data answers what a defined group of employers pays for matched jobs on a given date. The band answers what this company has chosen to pay, for which jobs and against which competitors. Those are different questions.
The common approach lets the tool's defaults answer the second question by accident. We would rather a company make four choices out loud for each family: The market, the position, the match and the date. The data then fills in the number. The result is bands that cost what the company intended and that a manager can explain to the person in the job.
When a band is challenged, by a candidate, an employee or a regulator, the strongest answer is never "that's what the data said". It is "this is the market we compete in and where we chose to sit in it".
Where we help
Our pay bands and structures work turns the market data you already have into bands for your levels: We set the market and position per family with you, check every job match, test the bands against current pay and write job ad range wording that matches them. The entry point is our Pay Band Sprint.* If you would like to talk it through first, Book a free call.
*We don't provide market data. We use the data you have or buy. No data yet? We help you choose a survey or tool that fits your size and budget.
Sources
- CBIZ, What conflicting survey data may reveal about compensation philosophy (23 April 2026), checked 2026-10-20
- WorldatWork, Workspan Daily: Employers' 2027 pay budget projections point to stability (16 July 2026), checked 2026-10-20
- WorldatWork, press release: 2025 salary increase budgets moderate (31 July 2025), checked 2026-10-20
- US Department of Justice and Federal Trade Commission, Antitrust Guidelines for Business Activities Affecting Workers (January 2025), checked 2026-10-20
- Washington L&I, Administrative Policy ES.E.2, Job Posting Requirements (issued 3 December 2025), checked 2026-10-20
- California Labor Commissioner, Equal Pay Act FAQ (questions 77 and 83), checked 2026-10-20
- Massachusetts Attorney General's Office, Pay Transparency in Massachusetts (FAQ 9), checked 2026-10-20
- Illinois Department of Labor, Equal Pay Act salary transparency FAQ (question 13), checked 2026-10-20