The filing itself · Wage year 2026-27

Using a private wage survey instead of the OES figure

An employer isn’t locked to the published tables. It’s locked to a set of conditions, and most salary data on the market fails at least one of them.

Read this firstThis is a computed reading of published federal data and published federal guidance, not legal advice. Only the Department of Labor issues prevailing wage determinations, and only the Department of Labor decides whether an obligation has been met. Where a rule below is our reading rather than a verbatim quotation, the text says so.

The published wage tables are the default, not the only permitted source. The regulations contemplate wages set by a collective bargaining agreement, wages set under the Davis-Bacon or Service Contract Acts for certain work, and wages supported by an independent authoritative survey. The last is the one employers reach for when the published figure looks wrong for the actual job.

The reasons to reach for it are usually real. The published figure may be imputed from a broader geography rather than surveyed locally, a problem described in What GeoLvl means, and when a wage figure is imputed. Or the occupation code may be broad enough that the survey population includes work quite unlike the job in question.

What a survey has to satisfy

The conditions below are our reading of 20 CFR 656.40 for permanent labor certification and 20 CFR 655.731 for Labor Condition Applications. They’re a summary, not a quotation, and the regulations themselves are what governs. Read them before relying on any survey.

• Independence. The survey has to be conducted by someone other than the employer relying on it, and it has to include employers beyond the requester.
• Occupational fit. It has to cover the occupation in question rather than a job title that happens to sound similar, and it should reach across industries employing that occupation rather than one narrow slice.
• Geographic fit. It has to represent the area of intended employment. A national figure isn’t a survey of a metropolitan area.
• Currency. There are limits on how old the published survey and the underlying collected data may be. Both dates matter and both have to be documented.
• A stated methodology. How many workers, how they were selected, what the wage measure is, and how it was computed. A figure without a method behind it can’t be evaluated by anyone.
• The right statistic. The regulations specify which measure of central tendency is used. Picking a percentile because it produces a better number isn’t available.

Why most salary data fails

Commercial compensation products are built for a different purpose: helping employers price offers competitively. That purpose rewards recency, narrow peer groups and self-reported data, and each of those is a problem here. Data scraped from job postings isn’t a survey of wages paid. A tool that lets the user choose the peer set isn’t independent of the requester. A figure with no published methodology can’t be documented.

None of that makes such data useless. It makes it evidence about the market rather than a substitute for the prevailing wage source, and those are different jobs.

What it costs to do it this way

The published figure requires no defense. A survey figure requires the survey, its methodology, the reasoning connecting it to this job and this area, and all of it retained where it can be produced later. For a Labor Condition Application that means the public access file, described in What a public access file has to contain. For a permanent case it means the audit file. The work is real, and it’s worth doing only where the published figure is genuinely unrepresentative.

Before going down that road, check what the published tables actually say for the occupation and area in question. For Software Developers in New York-Newark-Jersey City, NY-NJ, wage year 2026-27, Level I is $109,845 and Level IV is $193,461 a year, and a Level I figure that looks unrealistically low is frequently a level question rather than a survey question. The distinction is worth settling first, using How the four wage levels are determined.

Computed from ALC_Export.csv in OFLC_Wages_2026-27.zip, wage year 2026-27.

The alternative route, and the one that produces a figure the department itself stands behind, is The prevailing wage determination request, Form ETA-9141. An employer offering a survey can submit it with that request rather than relying on it unilaterally.

More in the filing itself

• Reading an LCA: the fields that matter. The eight or so fields on Form ETA-9035 that carry the wage promise, and how to check each one.
• What an LCA actually certifies, and what it doesn’t. The narrow thing certification establishes, and the four things it leaves entirely open.
• The prevailing wage determination request, Form ETA-9141. The form that gets you a figure the department issued, when it’s required, and what it buys.
• Full time versus part time, and how the wage is stated. Hourly to annual, how part-time hours are stated, and what’s owed when the work stops.

All 24 guides are indexed at Guides. Wage figures throughout are read from ALC_Export.csv, wage year 2026-27.