Compensation surveys
Know what the market actually pays
Everything that goes into running a compensation survey: the types, the trade-offs, the templates, and the mistakes that quietly ruin the data.
What is a compensation survey?
A compensation survey collects pay and benefits data from a defined set of employers, then reports it as aggregated statistics so each participant can see where its own pay sits against the market. Employers use the results to set ranges, defend pay decisions, and track how fast their market is moving.
The useful ones share four properties: jobs matched on content rather than title, a clear definition of what is inside each number, percentiles rather than a lone average, and an effective date. Our complete methodology guide works through each one.
Who runs them
Three kinds of sponsor, three different goals
Before choosing a survey type it helps to know which seat you are in, because the same instrument serves very different purposes depending on who is fielding it.
A single employer
You want to know whether your pay is competitive and internally consistent. You need data on employers you compete with, which you cannot collect yourself without asking competitors to hand it over. Most employers in this seat either buy syndicated data or join a survey someone else convenes.
An economic development authority or chamber
You are convening the survey for a region. Your members cannot survey each other, but you can, and the resulting benchmark becomes a public good or a paid product for the county. Participation is the hard part, and it runs on the trust your organization already has.
A trade or professional association
You are convening for an industry rather than a geography. Your members compete nationally for specialized roles, so an industry cut answers questions a regional survey cannot, and the annual report becomes one of the more defensible reasons to keep paying dues.
The second and third groups are who we work with most. If you are convening on behalf of others, the pages for economic developers, chambers of commerce, and industry associations go into how these programs get funded and run.
The four types
Which kind of survey do you actually need?
Most confusion about compensation surveys comes from treating four different instruments as one thing.
Industry surveys
Cover one industry across many employers. The right choice when the work is specialized enough that an outside benchmark would not be comparable, or when the people you hire move between employers in your sector rather than between employers in your town.
- Manufacturing, healthcare, nonprofit, and more
- Usually sponsored by an association
Geographic surveys
Cover one region across industries. The dominant model for economic development authorities and chambers, because employers in a county compete with each other for the same labor regardless of sector. A warehouse and a hospital compete for the same entry-level candidate.
- County, metro, or multi-county
- Sponsored by an EDA, chamber, or state agency
Occupational surveys
Cover one job family wherever it appears. Useful for roles that clear at a national rate rather than a local one, which is more common the more specialized the role. A staff accountant is local; a tax attorney with a niche specialty is not.
- One job family, many industries
- Broader market cut
Executive and board surveys
Cover senior leadership against a named peer group. A separate instrument because base pay is often the smallest component and the cohort is small enough that confidentiality drives the design rather than being applied at the end.
- Total compensation unbundled
- Higher suppression threshold
Formal versus informal
An informal survey is a few calls to peers. It is fast, and it is not evidence. A formal survey uses standard occupation codes, a fixed instrument, and stated statistics, so the result can be repeated next year and compared. The difference shows up the moment someone asks you to defend a number.
- Repeatable year over year
- Defensible to a board
Wages versus benefits
These behave differently and are usually separate sections or separate surveys. Wage data is about levels. Benefits data is about prevalence and design: which plans exist, how generous the employer contribution is, how time off accrues, how a match is structured.
- 174-question benefits template included
- Structured formats, not free text
The market cut
Drawing the boundary around your market
Choosing the survey type settles what kind of instrument you are building. Choosing the market cut settles who is in it, and that decision moves your numbers more than almost anything else you will do. Three questions get you most of the way there.
Where do candidates actually come from?
Not where you advertise, where people accept from. For most hourly and administrative roles that radius is smaller than it feels, often a single county or a short commute. For senior and specialized roles it can be a whole region or the country. Draw the boundary around the second group and you will overstate what you need to pay the first.
Does industry change the work?
An accountant is largely an accountant anywhere, so an industry cut adds little. A maintenance technician in a food plant and one in a machine shop do measurably different jobs under the same title. Cut by industry where the work genuinely differs, and skip it where you would only be shrinking your sample for no gain.
Will the cut survive suppression?
Every additional dimension divides your respondents into smaller groups, and small groups get redacted. County plus industry plus career level is three cuts deep, which is often one more than a regional cohort can support. Decide which cut you would give up first, before the report tells you.
Most regional programs settle on geography as the primary cut with industry as a filter, which keeps the sample intact while still letting a manufacturer compare against other manufacturers where the numbers support it.
By industry
Compensation surveys in your sector
Each industry breaks a survey in a different place. These pages cover the roles, the pay components, and the confidentiality problem specific to each.
Manufacturing wage survey
Benchmark production, skilled trades, and supervisory pay against the plants you actually compete with for labor.
Healthcare compensation survey
Benchmark nursing, clinical, and administrative pay across the employers competing for the same clinicians you are.
Nonprofit compensation survey
Benchmark program, development, finance, and leadership pay against comparable organizations, with data a board will accept.
Executive compensation survey
Benchmark C-suite and board pay across a defined peer group, with the confidentiality controls the data demands.
The decision
Buy syndicated data, or run your own survey?
This is the fork most organizations hit first. Syndicated providers sell access to large national datasets built from thousands of employers. They are comprehensive, they arrive without you doing any work, and for broadly-held corporate roles they are often the right answer.
They are the wrong answer when the market you care about is narrower than the dataset. You do not lose a maintenance technician to the national average. You lose them to the employer twenty minutes away. A national median can be perfectly accurate and still tell you nothing actionable about a county labor market, and the metro-level cuts that would help are usually priced as an add-on.
Running your own survey inverts the trade. The dataset is smaller, and it describes exactly the employers you compete with. It also compounds in a way purchased data does not: link this year to last year and you own a trend line for your region that nobody else has.
What it takes
The parts that usually go wrong
Compensation surveys fail quietly. Nobody spots a bad benchmark the way they spot a broken form, because a number that looks plausible still gets used, cited, and built into a budget. Each of the failures below costs almost nothing to prevent while you are still designing the instrument, and cannot be repaired once the responses are in.
- Jobs matched on content, not title. A coordinator at one organization does the work of a manager at another. Standard SOC codes with career-level ladders are how you keep the comparison honest.
- A stated definition of each number. Base only, base plus incentive, or total cash. A report that never says which has left every reader to guess, and they will guess differently.
- Spread, not just a midpoint. Almost every job pays across a band. Report the 10th, 25th, 50th, 75th, and 90th percentiles, because the width of the band is often more informative than its centre.
- Confidentiality participants can verify. Safe-harbor suppression with a threshold you set, plus per-occupation protection. If participants doubt it, they shade their answers and the data degrades without anyone noticing.
- An effective date on every figure. All data ages from the moment it is collected. You cannot combine or compare numbers without knowing when each was true.
- A path to next year. Link this cycle to the last one so returning respondents pre-fill and you get a trend instead of a series of disconnected snapshots.
Effort and timeline
What running one actually involves
The software is rarely the constraint. Planning decisions and participant recruitment are what set the calendar, and both are front-loaded.
Phase 1
Design and scoping
Agreeing the job list, the market cut, and the confidentiality threshold. This is the phase worth slowing down for, because every later problem traces back to a decision made here. Every license includes a pre-launch review of each question, and a deeper Ph.D. methodology review is available as an add-on.
Phase 2
Building the roster
Assembling and deduplicating the invite list. If you ran a survey before, we migrate it at no charge, which means last year’s roster and responses come with you rather than being retyped.
Phase 3
Fielding
Usually two to three weeks in the field, long enough for a respondent to pull payroll data and come back. Autosave, save-and-exit, and scheduled reminders do most of the work of getting responses over the line.
Phase 4
Compiling and release
Results compile automatically the moment the survey closes, into both an internal view and a safe-harbored audience view, then wait on your release. You choose who sees what and when, and you can preview each audience view before anything goes out.
Participation
The constraint is rarely the software
A compensation survey with twelve responses is not a small version of a survey with sixty. It is a different object, because suppression removes most of what twelve employers report and what survives cannot be sliced. Participation is what decides whether you publish a benchmark or a page of redactions, and it is worth more attention than any other part of the project.
Who asks matters more than how you ask
An invitation from a chamber, association, or county authority that an employer already belongs to performs very differently from the same invitation sent by a vendor they have never heard of. This is the main reason convened surveys work at all, and it is why we recommend the sponsoring organization stay visibly in front of the outreach even when we are running everything behind it.
The report is the incentive
Employers participate to see the results. Make that trade explicit: participants get the full benchmark, non-participants get less or pay for it. Gating results is not a hard sell, because the people you are asking already understand that the data only exists if enough of them contribute.
State the confidentiality rule up front
Telling participants the suppression threshold before they answer converts more of them than any reminder will. It also improves the answers you do get, because a respondent who trusts the protection submits real figures instead of rounded ones.
Remove the reasons to stop
Most abandonment is mechanical rather than motivational. Compensation data requires looking something up, so anything that forces one sitting costs you completions. Autosave, save-and-exit, prefilled prior answers, and a short call with a stuck participant recover responses that no reminder email will. The operational checklist covers the fielding tactics in order.
Free templates
Start from something that works
The instruments we use with clients, available to download and adapt.
Benefits survey questions
The benefit categories worth covering and how to word them, drawn from our 174-question template.
Benefits survey template
A full benefits instrument covering medical, retirement, PTO, leave, and development, ready to edit.
Salary benchmarking template
The wage grid: occupation down the rows, career level across, percentile bands in each cell.
How does Sensible Surveys compare?
A quick side-by-side against the alternatives most prospects evaluate. Competitor details reflect publicly listed offerings as of 2026; check their sites for the latest.
| Sensible Surveys | Culpepper | SurveyMonkey / Typeform | Alchemer | Hire-a-consultant | |
|---|---|---|---|---|---|
| Pre-built compensation & benefits templates | ✓ | ✓ | – | – | Custom every time |
| Built for Economic Developers & Associations | ✓ | – | – | – | Sometimes |
| Done-For-You option | ✓ | Their own industry surveys only | – | Setup services only | ✓ |
| Pre-launch survey design review | ✓ | – | – | – | Depends on consultant |
| Anonymized analytics (Safe Harbor) | ✓ | ✓ | – | – | Depends on consultant |
| Year-over-year comparability | ✓ | ✓ | – | – | Hard, different consultant each year |
| Local data focus (county / regional) | ✓ | – | – | – | Sometimes |
| Pricing model | Annual platform license | Annual data license | Per-seat subscription | Per-seat subscription | Project-based, $20–80k typical |
Common questions
How many participants does a compensation survey need?
There is no single number. What matters is whether the roles you care about clear the confidentiality threshold you set, since any question answered by too few distinct employers is redacted. A survey with thirty employers can be excellent for a narrow role list and useless for a wide one. We model your expected cohort against the threshold before you field.
What questions should a compensation survey ask?
For wages: the job on a standard occupation code, the career level, the pay range, and a clear statement of what is inside the number. For benefits: plan types, employer contribution, and accrual rules rather than free text. Our benefits survey questions page covers the categories in detail.
How often should we run one?
Annually is the common cadence, because year-over-year comparison is where the data becomes genuinely useful. The second cycle is also easier than the first: returning respondents are matched to their prior records and their answers pre-fill, so they review and update instead of starting over.
Can participants identify each other in the results?
Not if suppression is set up correctly. Any question answered by fewer than N distinct businesses is redacted automatically, individual occupations are protected separately, and only questions you designate as filter axes can slice the data, which stops anyone from narrowing filters down to a single employer.
Should we buy salary data instead of running a survey?
If your roles are broadly held and your market is national, purchased data is often faster and perfectly adequate. If your market is a county, a metro, or a specialized industry, a survey scoped to the employers you actually compete with will answer questions a national dataset cannot, and it compounds each year rather than expiring with the subscription.
How long does it take to run a survey?
Fielding is typically two to three weeks. The planning and recruitment ahead of it usually take longer than the survey itself, which is why we start with working sessions on the job list and the participant roster rather than with the software.
What does it cost to run a compensation survey?
Running it yourself on the platform is a flat annual license that includes up to three surveys a year. Having our team run it is quoted per survey based on audience size. See pricing.
The guides
Four pages covering the full arc, from the vocabulary through publishing the report. Start wherever you are.
What is a compensation survey?
The definitional version: the vocabulary you need before your first planning meeting, and the questions to ask about any survey you are handed.
How to run a wage and benefits survey
The complete methodology guide. Instrument design, job matching, fielding, and publishing results that hold up.
The compensation and benefits survey checklist
Ten steps for buying into a survey and using what comes back: choosing a provider, mapping roles, and turning percentiles into pay bands.
Reading your survey results
What belongs in the report, how to read percentiles honestly, and the results that should not drive a pay decision.
Complex surveys to send and don’t know where to start?
Book a 20-minute demo. We’ll show you the system, talk through your goals, and tell you whether we’re a fit. No pressure.