Job matching: Where benchmarks go wrong
Most bad benchmarks are bad matches, not bad data. The survey can only price the job you say you have.
Short answer: Job matching is where most salary benchmarking goes wrong. A survey prices the job it describes, so a match one level off or built on the title moves the number more than switching data sources would. Match the work the job does today, not the person or the title. Record how strong each match is and have one person review all the matches together.
Half the company underpaid, nobody leaving
Picture a 230-person logistics company in Indianapolis that runs warehouses and freight brokerage for regional retailers. It buys its first salary survey in the spring. The HR coordinator has two weeks to match about 70 jobs to the survey's catalog before the data is due.
The results come back in June. On paper, close to half the company is paid below the market median. Warehouse team leads look 15% under market. The head of finance looks a third under. Customer service looks fine. The finance lead asks the obvious question: If half the company is underpaid, why is turnover the lowest it has been in years?
A closer look answers it. The warehouse "team leads" were matched to the survey's supervisor job, which hires, disciplines and plans the shift. Here the team leads assign work and step in on the floor. The shift manager does the rest. The head of finance was matched to a finance director job that in the survey's sample runs treasury, tax and a team of twenty. The head of finance here runs a team of four. Three account coordinators were matched to the survey's clerk job, from a job description written eight years ago. Two logistics analysts, who do much the same customer reporting, were matched to an analyst job one level up. The coordinators are women. The analysts are men.
The data was fine. Almost every number that looked wrong traced back to a match.
This is a composite of a common situation at companies this size, not a client. The details are illustrative. We see the pattern often because matching looks like admin, so it gets the least time of any step in benchmarking.
Why job matching is hard
Job matching means deciding which job in a salary survey does the same work as a job in your company, at the same level. The survey then reports what other employers pay for that job. Everything that follows, from the band to the offer to the pay equity review, inherits the match.
Two sound goals pull against each other. The first is speed and coverage. A survey is worth more the more of your jobs it prices and the matching has to be done by a deadline, usually by someone with other work. The second is accuracy. A good match needs someone to read the survey's description and compare it with what the job actually involves today. That takes time per job and judgment that may not sit with the person doing it.
The usual standard is not perfection. A common rule of thumb treats a survey job as a match when about 70% or more of the responsibilities are the same, judged from the job's summary and main duties, not the title [1]. That leaves room for judgment on every match. It also means a weak match and a strong match can produce numbers that look equally precise.
Titles make it harder still. The federal statistics agency that levels jobs for its own pay survey notes that what one employer calls a supervisor, another calls a working lead and a third a senior employee, so it levels them by responsibility "regardless of titles" [2]. Title inflation adds to the problem. In an August 2026 hiring survey in India, 57% of professionals said their organization uses senior-sounding or enhanced titles to attract or keep people [6]. In a US survey of 1,000 workers, 57% said colleagues with the same title had significantly different pay or responsibilities [7].
Finally, the person matching is rarely neutral. A manager knows the job best and also wants the number to come out a certain way. Our guide on pricing a role with no market match covers what happens when no survey job fits. This guide is about the more common case: A match exists and gets chosen badly.
Four kinds of match and what each can carry
Not every match deserves the same weight. This table compares the four kinds most companies end up with. It is illustrative, not a matching guide.
| Kind of match | What it means | What it can carry | The usual risk |
|---|---|---|---|
| Strong | Most of the work and the level are the same as the survey job | A band midpoint, offers and pay equity comparisons | Going stale as the job or the survey changes |
| Partial | Much of the work is the same, but a real part differs | A band, with the gap noted and internal comparisons given weight | Being treated as strong once the number is in a spreadsheet |
| Level-off | The right kind of job, at a scope above or below yours | Direction only, until the level is settled | Moving the number more than any other single error |
| No match | No survey job covers most of the work | A blend of close jobs, openly labeled | A forced match that hides the guess |
The first row is where a company wants most of its important jobs to sit. The traps below are mostly about jobs that are really in the second or third row but get treated as if they were in the first.
Five traps in job matching
None of these comes from carelessness. Each is a reasonable shortcut under a deadline. The cost shows up months later, as a band nobody trusts or a gap nobody can explain.
"Match the person, not the job"
What it looks like: The match is chosen by thinking of the person in the role. A strong analyst is matched to the senior analyst job because "she operates at that level". A new hire in the same role is matched one level lower because he is still learning.
Why smart teams fall into it: Managers know their people better than they know their jobs. Describing what a strong performer does feels like describing the job.
What it costs: The benchmark now measures performance, not the market for the work. Two people doing the same job get different market rates, so the band for one job has two answers. Pay decisions that should come from a performance review get baked into the market data, where nobody looks for them. When the strong performer leaves, her replacement inherits a market rate set for her. Equal pay law also looks at the job, not the person's qualities. US regulations say that equal pay does not depend on job titles or classifications but on what the job actually requires and what the employee actually does [3]. A level set by how much a manager rates someone is hard to defend as a fact about the job.
"Our people do more, so match them up"
What it looks like: Jobs at a growing company cover more ground than the survey's descriptions, so they are matched to the job one level up. The head of finance who "does everything" is matched to a finance director. The marketing manager who runs brand, events and the website is matched to a senior manager.
Why smart teams fall into it: It is often true that people at a company of 150 cover more ground. A survey job that does one slice of the work looks too small.
What it costs: Breadth is not level. A survey's levels usually describe depth, complexity and the size of the decisions, not the number of topics. The head of finance at a 150-person company covers more ground than a finance director at a large one. The director usually runs a bigger team, a larger budget and harder problems in each area. When a job does not fully meet the description of a level, the federal leveling guide uses the next level down, not the next level up [2]. Matching up has the opposite rule. A level-off match moves the benchmark more than most percentile debates. It also tells staff the company is paying below market when it may not be, which is how a company with low turnover talks itself into a raise budget it did not need. Our guide on setting bands from benchmark data shows how one level of error carries through to the midpoint.
"Match it from the job description on file"
What it looks like: Matching is done from the job descriptions in the HR system, many of them written years ago or for a job ad.
Why smart teams fall into it: It is the only written record of the job. Interviewing managers about 70 jobs feels impossible inside a two-week survey window.
What it costs: A job description describes the job as someone wanted it to look on the day it was written. A description written for an ad is written to attract, with the title and duties pitched high. A description eight years old describes a job that has since grown or shrunk. Either way, the match is to a document, not to the work. Titles are the weakest part of any description. US regulations put it plainly: Job titles are often too general to give much guidance about the work [3]. Matching from paper is fine when the paper is current. When it is not, a short check with the manager on what the job does now catches most of the error in a few minutes per job.
"Let each manager match their own team"
What it looks like: To save time and use local knowledge, each department head matches their own jobs. HR collects the results and loads them.
Why smart teams fall into it: Managers know the work. Spreading the task means it gets done before the survey deadline.
What it costs: Each manager applies a different standard and some have a stake in the answer. A manager who wants a raise for the team matches up. A cautious manager matches down. The company ends up with jobs of similar size matched a level apart in different departments, which is the kind of gap that is hardest to explain later. The pattern is rarely neutral by sex. A 2026 experiment found that when the same job was described as mostly held by women, people suggested a lower starting salary for it, about $865 a year less. Men and women did so alike [8]. Matching carries the same risk: Support and administrative jobs, often held by women, are easier to match down to a clerk job than similar work held by men is to an analyst job. In the EU, pay structures must let jobs be compared on objective, gender-neutral criteria: Skills, effort, responsibility and working conditions [5]. Matches that differ by department and by who holds the job undermine that comparison before any pay is set.
"Match once and never look again"
What it looks like: The first year's matches are kept and reused. Each year the new survey data is loaded against the same job codes without anyone rereading the matches.
Why smart teams fall into it: Matching was hard work the first time. Reusing it makes the annual refresh quick and keeps the numbers comparable year to year.
What it costs: Jobs change and so do surveys. A job that took on a team, or lost one, is now matched to the wrong level. A survey that splits one job into two, or changes its level definitions, quietly moves every number tied to the old code. The strength of each match was usually never written down, so nobody can tell which of last year's matches were strong and which were the best available guess. When an employee questions their band, or a regulator asks how a pay difference was set, "the survey said so" is only as good as the match behind it. US guidance accepts market value as a reason for a pay difference only where the employer can show it assessed the market value of the job [4]. A match that nobody has looked at in three years is not much of an assessment.
What sound job matching looks like
Sound matching gives a company numbers it can explain job by job, a year later, to someone who was not in the room.
Levels come first. Matching works best when the company has already decided the level of each job on the work it does. Then matching is a check of two descriptions against each other, not a debate about how senior a person is. Our guide on mapping titles to levels covers that step and its people side.
The match is to the work as it is today. It rests on what the job does now, confirmed with someone who knows it, not on the title, the person in the job or an old description.
Breadth and level are kept apart. Wide-ranging jobs at a smaller company are matched on the depth and size of the work. Where a job spans two survey jobs, it is treated as a blend and labeled as one, not matched up.
Every match has a strength on record. Strong, partial or weak, with one line on what differs. Weak matches carry less weight in decisions and internal comparisons carry more.
One person reviews all the matches together. Managers can propose. Someone without a stake in any one department's answer looks at all the matches side by side, by level and by family, before any number is used.
Similar jobs get similar matches. Jobs of similar size in different departments land at the same survey level, whatever their titles. The check is done with an eye on who holds the jobs.
Matches are reread every year. When the survey changes its catalog or a job changes its scope, the match is reviewed, not carried forward by default.
The calls only you can make
Good matching narrows the argument. Some choices still belong to the company.
Which jobs need strong matches? A company cannot match every job well in the first year. Decide which jobs matter most, usually the largest groups and the hardest to hire. Spend the matching time there.
What do you do with a weak match? Use it with a label, blend it with others or price the job on internal comparisons instead. The answer depends on how much the job matters and how far off the match is.
Who owns the matches? Someone has to have the last word. That person needs enough authority to overrule a department head and enough distance to have no stake in the answer.
How much does a bigger job at your company count? If your head of finance really does the work of a director, that is a leveling question. Settle it in your structure, on purpose, not inside the survey submission.
Which survey do you match to? Matching is easier when the survey's catalog fits your jobs. Our guide on choosing a salary survey covers why coverage of your roles matters more than brand.
The data, the law and the numbers
- The usual match standard. A common rule of thumb treats a survey job as a match when about 70% or more of its responsibilities are the same, judged on the job summary and main duties, not solely on the title. The page we cite is a university's market pricing guide and is undated [1].
- Leveling by the work. The US Bureau of Labor Statistics levels jobs for its National Compensation Survey by four factors: Knowledge, job controls and complexity, contacts and physical environment. It classifies supervisors by their level of supervisory responsibility, regardless of titles. When a job does not meet the full description of a level, the next lowest level is used [2]. The guide was revised in May 2013 and is still the version the agency publishes.
- Title inflation. In Indeed's quarterly Hiring Tracker for India, reported in August 2026, 57% of professionals said their organization uses senior-sounding or enhanced titles to some or a great extent to attract or keep talent. Nearly six in ten said their title does not reflect their pay [6]. In a US survey of 1,000 workers fielded in August 2025 by a resume-building service, 57% said coworkers with the same title had significantly different pay or responsibilities and 39% had been given a more senior title without a pay increase [7]. These are self-reported survey figures.
- How jobs held by women are valued. A 2026 study in Research in Social Stratification and Mobility showed a nationally representative US sample the same job description with different shares of women in the field. The version described as 67% women drew suggested starting pay about $865 a year lower than the version described as 25% women. Men and women set pay the same way [8].
- United States, equal pay. Federal equal pay regulations say equal pay does not depend on job classifications or titles but on actual job requirements and performance and that titles are often too general to give much guidance [3]. Equal pay guidance accepts market value as a factor other than sex only where the employer shows it assessed the marketplace value of the job and the person's qualifications [4].
- European Union. Employers must have pay structures that allow work to be compared on objective, gender-neutral criteria agreed with workers' representatives where they exist, including skills, effort, responsibility and working conditions (Directive (EU) 2023/970, Article 4(4)) [5].
This is general information, not legal advice. Equal pay rules differ by state and country and change over time. Have an employment lawyer review how you document the market data behind pay decisions.
Our view
We think most bad benchmarks are bad matches, not bad data. Companies spend weeks choosing a survey and an afternoon matching jobs to it, when the matching decides more of the final number. A survey can only price the job you tell it you have.
Settle levels first, on the work. Match each job to what it does today, not to the title, the person or an old description. Keep breadth and level apart, so a busy job at a small company is not matched to a bigger one at a large company. Record how strong every match is. Have one person with no stake in the answer review all the matches together and reread them every year.
A weak match is not a failure. It is a fact about the job that the company should know. The failure is a weak match that looks like a strong one.
Where we help
Our market pricing work starts with the matches: Reviewing each job against the survey's descriptions, recording how strong each match is and checking that similar jobs land at the same level before any number is used.* See market pricing for how a project is scoped and priced. 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
- Oregon State University, Office of Human Resources, Market pricing steps (undated), checked 2026-10-29
- US Bureau of Labor Statistics, National Compensation Survey: Guide for Evaluating Your Firm's Jobs and Pay (revised May 2013), pages 3 to 5, checked 2026-10-29
- 29 CFR 1620.13(e), Equal work: What it means (Legal Information Institute), checked 2026-10-29
- US EEOC, Compliance Manual Section 10: Compensation Discrimination, 10-IV.F.2.g (market factors), checked 2026-10-26
- Directive (EU) 2023/970 on pay transparency, Article 4(4) (EUR-Lex), checked 2026-10-28
- Business Today, report on Indeed's quarterly Hiring Tracker, India (18 August 2026), checked 2026-10-29
- CPA Practice Advisor, report on the MyPerfectResume Title Inflation Report (12 May 2026; 1,000 US workers, 7 August 2025), checked 2026-10-29
- Earth.com, report on Taylor and colleagues, Research in Social Stratification and Mobility (18 July 2026), with the journal article, checked 2026-10-29