Which salary survey should you buy?
Coverage of your own jobs beats the brand name. Test a salary survey on your job list before you sign.
Short answer: The one that matches most of your important jobs, with enough comparable companies behind each match, under a license that fits how you will use the data. At 100 to 300 people that is rarely decided by brand or total participant count. Test each salary survey against your own job list before you buy.
The survey that didn't fit
Picture a 210-person medical device company in Minneapolis buying its first salary survey. Three sales teams pitch. One has the best-known name in the field, a large participant list and a polished online tool. The CFO likes the name. The company signs.
The first month is a disappointment. When the head of people matches the company's 74 jobs to the survey's descriptions, only about half find a match she trusts. The regulatory affairs and quality roles that matter most to the business either are not in the survey or are described for much larger companies. When she narrows the data to medical device companies of similar size in the Midwest, several cuts show an asterisk warning of a small sample. Some show nothing at all. The data's effective date is the previous April.
Then a consultant helping with the company's new pay bands asks to see the data. The head of people reads the license for the first time. It limits the data to the company's internal use.
None of this means the survey is poor. It is a good survey for a different company. Choosing one is less about the survey's quality than about its fit to your jobs, your peers and how you will use it.
This company is a composite of situations common at this size. It is not a client and the details are illustrative.
Why choosing a survey is harder than it looks
A salary survey is a dataset collected from participating employers, who match their jobs to the survey's job descriptions and submit pay for the people in them. The provider cleans and aggregates the results and sells access, often at a lower price or free to participants. Every survey trades three things against each other.
Breadth. A large survey covers many employers and many jobs. That makes it reliable for common roles and broad cuts. A survey's headline participant count can be impressive. One well-known salary budget survey drew 1,773 organizations in 2025 [1]. The count says little about how many of them match a specific role in a specific industry, size and location.
Depth. A specialist survey covers fewer employers in one industry or job area, often in more detail. It may be the only place where niche roles are matched well. It may also be thin outside that niche.
Effort. Participation usually earns access or a discount [1]. It also takes real work: Matching every job, cleaning data and answering queries, often in a narrow annual window. For a company with one person in HR, that work competes with everything else.
Underneath sits the tension between two good goals. The first is credibility: Data from a source finance and the board will trust. The second is fit: Data that actually describes your jobs and the companies you compete with for people. The best-known source is not always the best fit. A perfect fit from an unknown source can be hard to defend.
Timing adds one more trade-off. Surveys collect data in a window and publish months later, with an effective date set by the provider. A company whose pay review falls just before publication works with data that is a year older than it needs to be. The calendar of the survey should fit the calendar of the pay review, not the other way round.
Free public data does not remove the choice. US government occupational wage data is free and rigorous. It covers broad occupations, combines survey panels collected over three years and is aged to a single May reference date [2]. That makes it useful as a check and too broad and slow to set one company's ranges.
Five ways companies choose the wrong survey
Each of these is a sensible-sounding shortcut. Each leaves the company with data it cannot use for the decisions it bought it for.
"Pick the biggest name"
What it looks like: The company chooses the survey its CFO or board has heard of, or the one larger competitors use.
Why smart teams fall into it: A known name is easy to defend. Nobody gets questioned for buying the market leader.
What it costs: Brand tells you about the provider, not about your jobs. A survey built mostly from large employers may describe jobs at a scope your company does not have. Your most important roles may not be in it at all. The test that matters is how many of your own jobs find a match you would defend. The second test is how many comparable companies sit behind each one. That can be checked before signing. Most companies do not ask.
"More participants means better data for us"
What it looks like: The sales pitch leads with the total participant count and the company treats it as a measure of quality.
Why smart teams fall into it: Bigger samples are better in principle. A survey with thousands of employers sounds safer than one with hundreds.
What it costs: What counts is the size of the cut you will actually use: Your industry, your size band, your location and your job. A large survey can still have very few companies in that cell. Survey publishers know this and say so. One major budget survey marks figures that may rest on fewer than 30 organizations and hides any figure based on fewer than five [1]. A company that cuts the data finely enough to look like itself often ends up in exactly those cells. A smaller survey that is dense where you are can be worth more than a large one that is thin.
"We'll just buy it, not participate"
What it looks like: The company buys results without submitting its own data, to save the effort.
Why smart teams fall into it: Participation takes weeks of work at a time of year when HR is already stretched. Buying looks like the efficient choice.
What it costs: Participants often get results at a lower price or free. One major survey gives participating members a complimentary subscription [1]. The bigger loss is less visible. Participating forces the company to match every job carefully against the survey's descriptions, which is the same discipline that makes the data usable later. A company that only buys tends to match quickly when it needs a number, which is where most benchmarking errors start. Our guide on whether market data can set your salary bands looks at how weak matches distort bands.
"The survey will tell us where our jobs fit"
What it looks like: The company expects the survey's job catalog to sort out its structure. Jobs are matched by title. The level is whatever the survey's description seems closest to.
Why smart teams fall into it: Survey job catalogs are detailed and well written. It is tempting to let them stand in for a structure the company has not built.
What it costs: A survey can only price the jobs you bring to it. If your company has no clear levels, each match is a guess about scope. The guesses will not be consistent. Two people matching the same job on different days can land a level apart, which can move the benchmark by a large share of pay. Our guide on job families and levels explains why levels come first. A survey bought before levels exist is often used badly for a year.
Matching also needs a rule for partial matches. Many jobs at a company this size combine two survey jobs, or sit between two levels. Whether the company blends them, picks the closer one or leaves the job unmatched is a decision. Made once and written down, it keeps the data consistent. Made case by case, it lets the preferred answer creep in.
"We can share the numbers with whoever needs them"
What it looks like: Survey data is passed to consultants, pasted into board papers and quoted in conversations with peer companies.
Why smart teams fall into it: The company paid for the data. Using it freely feels like getting value from the purchase.
What it costs: Two separate limits apply. The first is the license. Survey terms commonly restrict data to the buyer's own organizational use and forbid reproduction or redistribution without permission [1]. Sharing data with an outside adviser may need the provider's consent. The second is competition law. US federal guidance once described a "safety zone" for wage surveys. The Justice Department withdrew it in 2023, calling the old statements overly permissive on information sharing [3]. The agencies' 2025 guidelines list sharing information about wages among companies that compete for workers as conduct that may violate antitrust law [4]. Their 2026 request for comment on guidance for competitor collaborations asks about information and data sharing again [5]. Talking pay numbers directly with a peer company is not the same as reading an aggregated survey. The difference matters. This is a question for your lawyer, not your survey provider.
The practical line most companies draw is simple. Pay data goes out only into surveys run by an independent party that aggregates it. It comes back only in aggregated form. Anything else, from a call between two heads of people to a shared spreadsheet among founders, gets checked first.
What a well-chosen survey looks like
No survey is right for every company. A good choice for yours has these qualities.
- It covers your jobs. Before signing, you have tested your own job list against the catalog and know what share of jobs find a match you would defend. You also know which important ones do not.
- Your cuts are dense enough to use. For the industry, size and location you will compare against, you know roughly how many organizations sit behind the numbers for your main roles.
- Its peer group matches the market you hire from. The companies in the data are ones you compete with for people, which may not be the ones you compete with for customers.
- Its method is clear. You know whether data is reported by employers or by individuals, how it is checked, what the effective date is and what is excluded.
- Its license fits your use. You know who inside and outside the company may see the data and in what form, before you need to share it.
- The effort is planned. If you participate, someone owns the submission and the matching, with time set aside in the survey window.
- It has a check. For the few fast-moving roles a broad survey prices poorly, a second source is used as a check, not a replacement. Our guide on which salary data to trust covers that choice.
- It is reviewed each year. The renewal is a decision, based on how well the data served last year's pay decisions, not an automatic payment.
One more quality is easy to overlook: The provider can explain a number. When a figure looks wrong, someone at the provider can say which companies and how many people sit behind it, within the limits of confidentiality. A provider that cannot answer that question is asking you to trust a number you cannot check.
None of this is a scoring sheet. Each quality is a question to put to a provider and to your own needs before signing.
The calls only you can make
Several decisions about market data depend on how the company wants to pay.
Which market you benchmark against. Industry peers, companies of your size or the employers your people leave for. Each gives a different answer. The survey can only answer the one you choose.
Whether to participate. Participation costs time and gives detail, discounts and discipline. Buying alone saves time and costs money and detail. At 100 to 300 people the time cost is real, so the decision should be made deliberately.
One survey or two. A single source keeps the structure consistent. Two sources widen coverage and invite choosing whichever number suits the moment. If you buy two, decide in advance which one leads for which jobs.
How much to spend. Survey costs vary widely. The right budget is the one that improves the decisions you make with the data, not the one that buys the most rows.
Who owns the matching. Matching decides more than the choice of survey. Whoever owns it needs the time, the job knowledge and the authority to say no to a manager's preferred match.
The data, the law and the numbers
- Sample flags. A major 2025 to 2026 salary budget survey of 1,773 organizations flags figures that may rest on fewer than 30 organizations and hides those based on fewer than five. Its data is for the participant's organizational use only [1].
- Free government data. US occupational wage estimates combine three years of survey panels, aged to a May reference date. They are reliable for broad occupations [2].
- Antitrust. The Justice Department withdrew its old safety-zone guidance in February 2023 [3]. The 2025 federal guidelines on business practices affecting workers name wage information sharing among competitors for workers as a risk [4]. In February 2026 the agencies sought comment on new guidance for competitor collaborations, including information and data sharing [5].
- Posted pay is not a substitute. Research from the Minneapolis Fed finds that posted wages run well above the pay of workers in government data, so job ad ranges are a weak replacement for a survey [6].
Our view
We think coverage of your own jobs beats brand name. The best survey for a company of 100 to 300 people is the one that matches most of its important roles, with enough comparable companies behind each, under a license that fits how it will use the data.
So we would test before buying. Take your job list to each provider and ask how many of your jobs they can match well and how many organizations sit behind those matches in the cuts you would use. The answers usually make the choice. A provider that will not run that test before you sign is telling you something about how the next year will go.
For most companies at this size, one broad survey that covers the core of the workforce, plus a narrow source for the few roles it prices poorly, is enough. More data rarely fixes a weak match. Better matching usually does.
We would argue hardest against two habits: Buying a survey before the company has levels and sharing survey numbers directly with peer companies. The first wastes the data. The second can create a legal problem that no survey is worth.
Where we help
Our market pricing work helps you test surveys against your own jobs before you buy, then matches your jobs carefully and brings the data to one date. 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
- WorldatWork, 2025-2026 Salary Budget Survey: Top-level results (published 15 July 2025), checked 2026-10-25
- US Bureau of Labor Statistics, Occupational Employment and Wage Statistics: Survey methods (page modified 15 May 2026), checked 2026-10-25
- US Department of Justice, Justice Department withdraws outdated enforcement policy statements (3 February 2023), checked 2026-10-25
- Federal Trade Commission, FTC and DOJ jointly issue antitrust guidelines on business practices that impact workers (16 January 2025), checked 2026-10-25
- Federal Trade Commission, FTC and DOJ seek public comment for guidance on business collaborations (23 February 2026), checked 2026-10-25
- Batra, Michaud and Mongey, Online job posts contain very little wage information, Federal Reserve Bank of Minneapolis Staff Report 690 (28 September 2026), checked 2026-10-25