Which market percentile to target?
The percentile is a promise to staff and a cost to finance. Pick it on purpose.
Short answer: For most companies the median is a sound market percentile to target, but only once you have said which market. The real decision is where to differ: Which roles get more, whether the target is base or total pay and how closely actual pay must follow it. A percentile nobody chose becomes a promise nobody can keep.
Two percentiles in one meeting
Picture a 220-person payments software company in Austin building its first salary structure. In the planning meeting the CFO proposes the 50th percentile for every role. It is the middle of the market, it is easy to defend and it is what the budget already assumes.
The VP of engineering wants the 75th. The company lost three senior engineering candidates last quarter, each to a higher offer. Two more engineers left for large tech employers in the spring. The head of people raises a different point. Nobody has agreed whose market they mean. The salary survey the company bought covers fintech companies across the US, while the engineers it keeps losing go to employers ten times its size.
After an hour they settle on the 60th "as a compromise". Nobody can say what the 60th promises, to whom or what it will cost in three years. Six months later a support lead asks why her offer sat below the midpoint of her range when the company says it "pays above market". Her manager has no answer that holds up.
The number they picked is not wrong. It was chosen as a number, not as a decision.
This company is a composite of situations common at this size. It is not a client and the details are illustrative.
Why choosing a market percentile is hard
A market percentile is a point in a survey's pay distribution for a matched job. At the 50th percentile, half the reported salaries are lower. At the 75th, three quarters are. A target percentile is the point a company says it aims to pay at.
Most companies start from the middle. In a 2025 survey of 133 Canadian employers, 44% targeted the 50th percentile for base pay and about a quarter aimed between the 50th and the 75th. About 2% aimed below the median and 7% had no defined position. Among employers with more than 500 staff, 65% targeted the median for base pay [1]. US surveys disagree on how strong the default is. A compensation advisory firm compared two 2026 reports. The one with more small and mid-sized respondents found 45% targeting the 50th and another 17% above market. The one weighted to larger employers found about 84% targeting the 50th for non-executive roles [2]. In an executive pay survey of 248 companies, 56% targeted the 50th for base salary and 27% aimed above it [3].
The difficulty is a real tension between two good goals. The first is cost control. Every point above the median applies to every role it covers, every year. It compounds, because merit increases are percentages of base pay. The second is attraction. At the median, half the market pays more by definition. Some candidates will always have a better offer. The people who make offers feel each loss.
A second tension sits inside the first. One percentile for every role is simple and looks fair. Different percentiles by role put money where the losses are, but they create groups who are paid differently and need a reason that survives being said aloud.
The percentile also hides most of the real choice. The same "50th" produces very different numbers depending on which employers are in the comparison and whether it applies to base salary, cash or total pay. Two companies can both say they pay at the median and be far apart.
And the choice no longer stays private. Posted pay ranges show candidates and staff roughly where a company sits. In the EU, the criteria used to set pay and pay progression must be easily accessible to workers [7]. A percentile that nobody chose on purpose is now one people can see.
Five ways companies get the percentile wrong
Each of these produces a target that sounds reasonable in a meeting and fails in practice.
"Pick the percentile, then work out the market"
What it looks like: The debate is about the 50th against the 75th. The comparison group is whatever the survey or tool shows by default.
Why smart teams fall into it: The percentile is the visible number and the one leaders argue about. The market feels like a technical setting someone else has already handled.
What it costs: The market moves pay more than the percentile does. The advisory firm's comparison makes the point directly: The 50th percentile of a broad market can be higher than the 75th percentile of a narrow one [2]. Most companies define their market by industry (83% in the 2025 survey) and region (71%). Only 40% define it by competition for the same roles [1]. That is how the Austin company came to compare its engineers with fintech peers while losing them to large tech employers. Moving to the 75th of the wrong market would cost more and still miss. Our guide on whether market data can set your salary bands covers how the comparison group shapes the structure itself.
"One percentile for everyone"
What it looks like: A single target applies to every role, from support to engineering to finance.
Why smart teams fall into it: It is simple, easy to explain and looks fair. Any difference between groups invites the question "why them?"
What it costs: A single target overpays where people are easy to find and underpays where they are scarce. The money goes evenly while the hiring problem is concentrated. The advisory firm reads the share of smaller employers targeting above market as mostly a response to pressure on specific roles, not a broad strategy [2]. Most companies already accept that markets differ by role: Only 19% use the same peer group for every level [1]. The honest version is a default position plus a short list of roles that are paid differently, each with a reason and a date to review it. That version has its own failure mode. Every leader argues that their team belongs on the list. Within two years everyone is "critical". A list that grows without a review is a single higher percentile reached slowly and without a decision.
"Aim high and the hiring problem goes away"
What it looks like: After a run of lost candidates, the company moves its target to the 65th or the 75th.
Why smart teams fall into it: Candidates who decline usually mention a higher offer. Pay is the quickest lever to pull and the easiest to measure.
What it costs: First, the cost stays. A higher target is easy to raise and very hard to lower, because lowering it means telling staff their pay will grow more slowly than promised. Second, the advantage fades. Research on executive pay found that when companies benchmark against peers and some pay above them, the raises spread through the peer groups and lift the whole market. A small, shifting group of companies paying at the top of their benchmarks accounted for a large share of the growth in executive pay [4]. If every employer aims above the median, the median rises. Third, the diagnosis is often wrong. Candidates who mention money may also be reacting to a slow process, a lower level than they expected, a weaker equity offer or a manager who did not impress them. Raising the target for every role fixes none of those. In the executive survey, the most common reasons for aiming above the median were stretch performance goals (36%) and competing with larger companies (28%). Only 10% cited tight labor markets, down from 27% a year earlier [3]. Reasons change faster than positions do.
"The target is where we pay"
What it looks like: The company says it pays at the median because its range midpoints are set there. Nobody checks where actual pay sits.
Why smart teams fall into it: The structure was built carefully, so it feels like the structure is the pay. Checking actual pay against the market takes work and may show something uncomfortable.
What it costs: Actual pay drifts from the target in predictable ways. New hires often start low in the range, while long-serving staff drift high. The market keeps moving after the survey is published: In a July 2026 survey of about 1,000 US organizations, three in four adjusted their salary structures every year, by an average of 2.6% [6]. A company whose midpoints were set at the median two years ago and never moved may now pay most of its newer staff well below it. The drift is rarely even. That matters more than the level. In a study of a large US retailer, quits responded strongly to pay relative to peers in the same company, much more than to pay relative to the market [5]. A lower target that is met evenly can hold people better than a higher one that is met for some and missed for others. Staff who hear "we pay at the 60th" and then see job ads for their role at higher figures stop believing anything else the company says about pay.
"Base salary tells the whole story"
What it looks like: The target is stated for base salary only, or for "total compensation" without saying what that includes.
Why smart teams fall into it: Base salary is the number surveys report most cleanly and the one candidates ask about first. Bonuses and equity feel too variable to include.
What it costs: Companies position different pay elements differently. In the 2025 survey, 44% targeted the median for base pay, 35% for total cash and 31% for total compensation [1]. In the executive survey, 56% targeted the median for base salary and 49% for total direct compensation [3]. A startup at the 40th percentile for base with generous equity may sit at the 70th for total pay on paper and well below it in what staff actually value. A company that says "we pay at the median" without saying of what leaves every candidate and employee to fill in the answer, usually with the element that looks worst. Staff discount bonuses they have not seen paid and equity they cannot sell. The target has to say which element carries the promise.
What a sound percentile choice looks like
A good target is less about the number than about what surrounds it. In our view it has these qualities.
- The market comes first. The company has said which employers it compares itself with for each group of roles, based on where it really hires from and loses people to.
- There is one default. A single position covers most roles and fits in one sentence.
- Exceptions are few, named and dated. Each role paid differently has a stated reason, such as documented hiring losses, plus a date when the exception ends or is reviewed.
- The element is stated. The target says whether it applies to base salary, total cash or total pay and how the other elements are meant to sit.
- Finance has seen the cost. The target has been costed over three years, not just this year, including how it compounds through merit increases.
- Actual pay is checked. At least once a year, someone compares where pay really sits against the target, for new hires and long-serving staff and for each job family.
- It survives being said aloud. A recruiter and a finance lead give the same answer to the question "where do we pay?"
None of this is a procedure. Each quality is a test a leadership team can put to the target it already has.
The most revealing test is the last one. Ask a recruiter, a manager and the finance lead separately where the company pays against the market. If they agree, the target is real. If they give three different answers, the company has three targets and none of them is written down.
The calls only you can make
The percentile is where a company decides how much it will spend to compete for people. Those decisions belong to leadership, not to the data.
Where to lead and where to follow. Leading the market everywhere is expensive. Following it everywhere ignores the roles you keep losing. Most companies need a default and a short list. Deciding what earns a place on the list is the real choice.
Whether above-market money goes to roles or to people. Paying a scarce role above the median rewards the market. Paying strong performers above it rewards contribution. Both are defensible. Doing both with one budget usually means doing neither well.
Which losses you accept. At the median, some candidates will always choose a higher offer. Deciding in advance which losses are acceptable stops each one from becoming a case for raising the target.
How much to say. Telling staff "we aim for the median of companies our size" builds trust if it is true and invites checking. Many companies state a position in their compensation philosophy and only some publish it.
How fast to follow the market. Moving the structure every year keeps the target honest and costs money every year. Moving it every two or three years saves money and lets actual pay fall behind the promise.
Who owns the number. A target set by HR alone gives way the first time a hiring manager pushes. One agreed by the CEO and the CFO can hold.
The numbers and the research
- Where companies aim. In a 2025 survey of 133 Canadian employers, 44% targeted the 50th percentile for base pay, 35% for total cash and 31% for total compensation. About a quarter aimed between the 50th and 75th, 2% below the median and 7% had no defined position. 65% of employers with more than 500 staff targeted the median for base pay [1].
- How they define the market. 98% used a defined peer group. Most defined it by industry (83%) and region (71%), 40% by competition for similar roles. 19% used one peer group for all levels [1].
- US surveys. One 2026 report found 45% targeting the 50th and 17% above market. Another, weighted to larger employers, found about 84% targeting the 50th for non-executive roles, as compared by a compensation advisory firm in April 2026 [2].
- Executives. In a survey of 248 companies fielded in August and September 2025, 56% targeted the 50th percentile for base salary and 27% above it. For total direct compensation the figures were 49% and 34% [3].
- Market movement. Three in four US organizations in a July 2026 survey adjust salary structures every year, by an average of 2.6% [6].
- Research. Peer benchmarking let raises at a few companies spread across executive pay [4]. At a large retailer, quits responded mainly to pay relative to colleagues rather than to the market [5].
- EU. The criteria used to set pay, pay levels and pay progression must be easily accessible to workers and must be objective and gender-neutral [7].
Our view
For most companies of 50 to 500 people, we think the right default is the median of the right market. Name the employers you really compete with first, then aim at the middle of them. Pay above it only for a short list of roles with documented hiring or retention losses, each with a review date.
We would state the target for base salary, describe how bonuses and equity are meant to sit and check actual pay against it every year. A modest target that is met evenly is worth more than an ambitious one that is met for some people and missed for others.
We would argue hardest against the compromise percentile: The 60th or the 65th picked because it sits between two leaders' opening positions. It costs more than the median, promises something nobody can define and leaves the real question unanswered. The percentile is a promise to staff. It should be made on purpose.
Where we help
Our market pricing work starts with the market: We agree with your leaders which employers you compete with for each job family, set a default position and the few exceptions and measure where your pay really sits against it. The starting point is a market pricing project, quoted within one business day. If a conversation would help 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
- Western Compensation & Benefits Consultants, Employer compensation philosophy, strategy and benchmarking survey (October 2025; 133 Canadian employers), peer groups and pay position strategy, checked 2026-10-25
- CBIZ, What conflicting survey data may reveal about compensation philosophy (23 April 2026; compares Payscale's 2026 Compensation Best Practices Report and WorldatWork's 2025 to 2026 Salary Budget Survey), checked 2026-10-25
- Pearl Meyer, Looking Ahead to Executive Pay Practices in 2026, executive summary (248 companies, August to September 2025), checked 2026-10-25
- DiPrete, Eirich and Pittinsky, Compensation benchmarking, leapfrogs and the surge in executive pay, American Journal of Sociology 115(6), 2010, checked 2026-10-25
- Dube, Giuliano and Leonard, Fairness and frictions: The impact of unequal raises on quit behavior, American Economic Review 109(2), 2019, checked 2026-10-25
- Mercer, Results of the July 2026 QuickPulse US Compensation Planning Survey (31 August 2026), checked 2026-10-24
- Directive (EU) 2023/970 on pay transparency, Article 6 (EUR-Lex), checked 2026-10-25