What is a salary survey, and how does it work?

The P50 team

Published October 2, 2026

Last updated October 5, 2026

A salary survey is a report of what employers actually pay for specific jobs. A third party collects pay data from many employers, combines it, and reports it back as market rates: for example, the median base salary for a staff accountant in your region. Employers use those rates to set pay ranges, check offers, and decide raises. Below we cover who runs surveys, how data gets in and out, and why most companies under 2,000 employees have never joined one.

What a salary survey is (and is not)

A salary survey is built from employer-reported data. The employer's HR or payroll team submits what it pays real people in real jobs. That is the key difference from salary websites, where workers type in their own pay and nobody checks it (see salary surveys vs salary websites). A survey is also not a list of what one company pays. The point is to pool pay from many employers so you can see the market.

Who runs salary surveys?

There are four main types of survey publishers.

Government. The U.S. Bureau of Labor Statistics runs the Occupational Employment and Wage Statistics (OEWS) program. It produces wage estimates for about 830 occupations, for the nation, each state, and metro and nonmetro areas, plus national estimates by industry (BLS OEWS). The sample covers about 1.1 million establishments over a three-year cycle (BLS OEWS sample design). It is free. The trade-off is that it reports by broad occupation code, not company job title, and it does not break pay out by employer size (BLS OEWS FAQ).

Consulting firms. The big surveys are run by compensation consulting firms. They publish thousands of job titles, cut the data many ways, and charge for the report. One large general-industry survey lists about 7,200 published positions and about 3,400 participating organizations (survey product page).

Industry associations. Trade groups for hospitals, credit unions, nonprofits, manufacturers, and local governments often run a survey for members. These are narrower but match your jobs well because the participants look like you.

Free surveys. A newer group collects employer-reported pay and gives the results back at no charge. P50 is one of these. Read the methods page of any free survey before you trust it.

How your data gets in

Almost every survey takes data in the same three steps.

Match your jobs to the survey's job catalog. The survey publishes a list of standard jobs with short descriptions. You pick the catalog job that fits each of your jobs best. Your "Accounting Specialist II" might match the survey's "Accountant, intermediate." This is where most of the work is. The big firms now offer a "proprietary job match algorithm and online job catalog" to speed it up (survey process page).
Submit pay for each matched job. You report base pay for each employee in that job, plus fields like location, bonus target, and sometimes job level. You do not send names. See how to build a pay file.
The survey checks your data. The publisher looks for outliers, mismatched jobs, and typos, and may send questions back before your data is accepted.

Surveys collect on a schedule. One large survey's 2026 collection window opened March 2 and closed April 24 (survey product page). Miss the window and you wait a year. See when do salary surveys open.

What comes back out

The survey pools everyone's data, then reports it by job. For each job you usually see:

The number of companies and employees reporting
Percentiles of base pay: the 25th, 50th (median), and 75th, sometimes the 10th and 90th
Averages for bonus and total cash
Cuts by location, industry, and sometimes company size

The percentiles are the heart of it. BLS defines the median as "the boundary between the highest paid 50 percent and the lowest paid 50 percent of workers in that occupation" (BLS OEWS FAQ). If you are not sure how to read them, start with how to read P25, P50, and P75, or look at our sample report.

What "data effective date" means

Every survey ties its numbers to a single date. Pay collected in the spring is reported as of that date, not the day you open the report. OEWS publishes with a May reference date, and wages from earlier panels are adjusted forward to that May using a model (BLS OEWS FAQ). The May 2025 estimates were released on May 15, 2026 (BLS OEWS).

This matters when you use the data. If the effective date is April 1 and you are setting pay in January, nine months have passed. Most teams "age" the numbers forward by an assumed market movement. We explain our approach on the methods page.

Why surveys cost what they cost

Running a survey is labor. Someone has to maintain the job catalog, recruit participants, chase late submissions, clean the data, and build the reports. The big firms recover that cost by selling the report, and they charge non-participants far more. For one large general-industry survey, the published 2026 price ranges from $2,600 to $7,100 per module for participants and $7,800 to $21,300 for non-participants, as of October 2026 (survey product page). Association surveys are usually cheaper but still cost money, and you often need to be a member.

Why companies under 2,000 employees often never join

Look at the gap. The SBA counts 19,688 U.S. businesses with 500 or more employees (2021 data) (SBA Office of Advocacy). The large survey above reports about 3,400 participating organizations (survey product page). Even among large employers, most are not in it. Below 500 employees, participation is rarer still.

Three things keep mid-size companies out:

Cost. A few thousand dollars for one survey is a hard line item when you have one HR generalist and no compensation budget.
Effort. Matching 80 or 200 jobs to a catalog takes days the first time, and you repeat some of it every year.
Matching homework. Many surveys expect you to learn their leveling system (what makes a job "senior" or "lead" in their terms) before you can match well.

So most mid-size employers price jobs from free government data, job postings, and gut feel. How to price a job without a salary survey covers the workarounds.

What "aggregated" and minimum-company rules mean for confidentiality

Survey reports are aggregated. The publisher never shows what one employer pays, only statistics (percentiles, averages, counts) computed across many employers. To make that hold, surveys enforce minimum-company rules: a statistic is published only if enough separate employers reported it, and no single employer dominates the number.

These rules come from antitrust practice. For years, the Department of Justice and the Federal Trade Commission described a "safety zone" for pay and price surveys among competitors. It had four parts: the survey is "managed by a third-party," the data is "more than 3 months old," there are "at least five providers reporting data upon which each disseminated statistic is based," and "no individual provider's data represents more than 25 percent on a weighted basis of that statistic" (DOJ and FTC 1996 policy statements, Statement 6).

As of October 2026, that safety zone is no longer official guidance. The DOJ withdrew the statements on February 3, 2023, calling them "overly permissive on certain subjects, such as information sharing" (DOJ press release), and the FTC followed on July 14, 2023 (FTC press release). Most reputable surveys still follow the old rules, and many go further. Ask any survey what its minimums are. Our data promise spells out ours, and can employers share salary data covers the legal side.

Survey types at a glance

Type	Who runs it	Cost to you	Job detail	Best for
Government (BLS OEWS)	Bureau of Labor Statistics	Free	About 830 occupation codes, not job titles	A quick sanity check by occupation and metro area
Consulting firm	Compensation consulting firms	Thousands of dollars per survey, much more if you do not participate	Thousands of job titles, many data cuts	Large employers with a compensation team
Industry association	Trade groups and member associations	Usually paid, often requires membership	Jobs specific to one industry	Hospitals, credit unions, nonprofits, local government
Free survey	Independent providers like P50	Free	Varies by provider, so check the methods page	Mid-size employers joining a survey for the first time

Where P50 fits

P50 is built for the last row of that table: employers who never joined a survey because the cost and matching work were too much. We collect employer-reported pay the same way the big surveys do, aggregate it with minimum-company rules, and give the results back for free. You can read how we count on the methods page and what we promise on the data page.

P50 is a free salary survey for employers. Registration for the 2027 survey is open now, and data collection runs March 1 to May 3, 2027. Register for free.