LCA vs H-1B Visa: What the Data Does and Doesn't Tell You

Is LCA data the same as H-1B visa data? No. A Labor Condition Application (LCA) is a wage-and-working-conditions attestation an employer files with the U.S. Department of Labor before it can petition for an H-1B worker. The visa petition itself is a separate step, decided by U.S. Citizenship and Immigration Services (USCIS), subject to an annual cap, and it is not part of this dataset. Everything on h1b.report is built from LCA disclosure files — so a "certified" record means the DOL cleared the wage attestation, not that anyone actually received a visa.

That distinction matters more than it sounds. It is the single most common source of bad H-1B statistics online. This article explains what the LCA data genuinely measures, why it is still enormously useful, and the five specific ways people misread it.

What the LCA data actually is

An LCA is the employer's sworn promise to the DOL: if we hire this worker, we will pay at or above the local prevailing wage for the occupation, and employing them will not harm conditions for U.S. workers. The form lists a job title mapped to a standard occupational code (SOC), a worksite, and an offered wage. Once the DOL certifies it, the employer can move on to the actual USCIS petition.

The disclosure data covers a lot of ground. This site normalizes roughly 9.3 million filings spanning fiscal years 2010 through 2026, from more than 152,000 employers, across 782 SOC occupations. That scale is exactly why the data is worth mining: it is the most comprehensive public window into what sponsoring employers say they intend to pay, by role, employer, and location.

You can slice it by occupation, by employer, by state, or by industry sector. The filings-over-time picture alone tells a clear growth story.

Why "certified" is not "approved"

Here is the first trap. At the LCA stage, denials are vanishingly rare — on the order of 0% to 0.6% across the largest sponsors. That is because the LCA is an attestation, not an adjudication of an individual's eligibility. The DOL checks that the paperwork is internally consistent and the wage clears the floor; it does not decide whether a person gets to work in the country.

The real gate is the USCIS petition that comes afterward, and this dataset never sees it. So a near-100% LCA certification rate does not mean a near-100% visa success rate. Certification means "the wage attestation passed," full stop.

The Microsoft example: withdrawn is not denied

The second trap is reading certification rates as approval scoreboards. Take a concrete case: Microsoft shows about 35.2% certified with 0.0% denied. At a glance that looks alarming — where did the other ~65% go?

They were certified-withdrawn, not rejected. Employers routinely file LCAs they later pull: a role gets restructured, a candidate takes another offer, a duplicate is filed across worksites, or hiring plans change. A withdrawn LCA is an administrative unwind, not a government "no." Reading withdrawals as denials turns an ordinary paperwork pattern into a false scandal. When you compare sponsors, look at what each status actually means before ranking anyone.

The wage on the form is a floor, not a paycheck

The third trap is treating the offered wage as take-home pay. The wage on an LCA is the offered or prevailing wage tied to the role and location — a compliance floor, disclosed as an annual figure. It typically excludes bonuses, stock, and other compensation, and it is not adjusted for taxes or cost of living. For big tech roles where equity is a large share of total pay, the LCA wage can sit well below what someone actually earns; for other roles it may track close to reality.

Used carefully, these wages are still a powerful benchmark, especially in aggregate. The highest-paying occupations in the data are a good example of a ranking that holds up precisely because it compares like with like.

Fiscal year 2026 is only partly in

The fourth trap is trend-chasing on incomplete data. FY2026 is partial — only about two quarters have been reported so far. Any year-over-year comparison that ends on FY2026 will understate the total simply because the year is not finished. When you read a filings chart, treat the most recent bar as a work in progress, not a decline.

The five guardrails, in one place

Keep these straight and you will read every H-1B chart on this site — and elsewhere — correctly:

• LCA ≠ visa. An LCA is a DOL wage attestation. The H-1B petition is a separate, capped USCIS decision this data never records.
• Certification rate ≠ USCIS approval. LCA denials are near-zero (~0–0.6%) because the LCA is not where eligibility is judged.
• Withdrawn ≠ denied. Microsoft's ~35.2% certified / 0.0% denied is mostly certified-withdrawn filings, not rejections.
• Offered wage ≠ take-home. The listed wage is a prevailing-wage floor, excluding bonuses, equity, taxes, and cost-of-living.
• FY2026 is partial. About two quarters are in; the latest year is incomplete by construction.

So what is the data good for?

Plenty — once you respect its boundaries. Because every sponsoring employer must file, the LCA record is an unusually honest map of demand: which employers are staffing up, in which occupations, in which states, and at what wage floors. It will not tell you how many people got visas, and it will not tell you anyone's real paycheck. But it will tell you, at national scale, how the sponsored labor market is priced and where it is concentrated.

If you are weighing a job offer, use the offered wage as a floor to sanity-check against — not a promise. If you are researching a sponsor, distinguish certified, withdrawn, and denied before you draw conclusions. And if you are tracking trends, mind the partial year. For the full methodology behind how these filings are cleaned and normalized, see our methodology page; for definitions of individual terms, the glossary and FAQ go deeper.

The data is trustworthy. It just answers a narrower question than its headline suggests — and knowing which question is the whole game.