Which Workforce Data Should You Use? OEWS vs QWI vs O*NET
The most common workforce-data mistake is not using bad data. It is asking a good dataset to answer the wrong question.
OEWS, QWI, and ONET answer different questions. Their value increases when the evidence remains separate and the decision boundary is explicit.*
Suppose a regional partnership is considering a new advanced-manufacturing training program. The team may want to know:
- How many people work in the relevant occupations?
- What do those jobs pay?
- Is the local industry expanding, replacing workers, or experiencing high turnover?
- What tasks and skills define the occupations?
- Are employers actually struggling to fill these roles now?
No single public dataset answers all five questions. Treating several sources as interchangeable produces a confident-looking analysis with weak foundations. A stronger approach starts by assigning each source a specific job.
OEWS: the occupation-and-wage view
Section titled “OEWS: the occupation-and-wage view”The Bureau of Labor Statistics Occupational Employment and Wage Statistics program produces employment and wage estimates for hundreds of occupations. Estimates are available nationally, by state, and for metropolitan and nonmetropolitan areas. OEWS also provides national industry-specific estimates.
OEWS is useful when the decision requires a structured view of occupational scale and compensation. It can help answer questions such as:
- How large is this occupation in the state or metro area?
- What are the mean and percentile wage estimates?
- Which occupations are relatively concentrated in a geography?
- How different are wages between locations?
But OEWS does not directly tell you whether a particular employer has vacancies, whether a training provider can deliver the required curriculum, or whether demand will persist after a new program launches. It measures occupational employment and wages; it is not a complete demand forecast.
QWI: the industry-and-worker-flow view
Section titled “QWI: the industry-and-worker-flow view”The Census Bureau Quarterly Workforce Indicators provide measures such as employment, earnings, hires, separations, job creation and job destruction. QWI can be examined by geography, industry, firm characteristics and worker demographics. It is available for states and several sub-state geographies, including counties and workforce areas.
QWI becomes valuable when the decision is about workforce movement rather than a static employment count. It can help teams investigate:
- Is employment in the target industry growing or contracting?
- Are hires being offset by separations?
- Is turnover persistent?
- How do earnings or employment patterns differ across locations?
- Which worker groups participate in the industry?
The important limitation is equally clear: QWI is primarily organized around industries and worker flows, not detailed occupations. An industry may employ many different occupations, and the same occupation may appear across many industries. QWI should not be silently converted into occupation-level demand.
O*NET: the work-and-skill view
Section titled “O*NET: the work-and-skill view”O*NET describes occupations through tasks, skills, knowledge, abilities, work activities, work context and other worker and job characteristics. It provides a structured language for understanding what work normally involves.
O*NET is useful when a team needs to:
- compare the task content of occupations;
- identify relevant skills and knowledge areas;
- distinguish between similar occupational candidates;
- connect a job title to a standardized occupational framework;
- review whether proposed training content aligns with the work.
O*NET does not tell you how many current openings exist in a specific county, what a local employer will pay, or whether one employer uses the occupation in the standard way. It is an occupational reference—not a live local demand feed.
The three-source decision pattern
Section titled “The three-source decision pattern”A disciplined workforce assessment keeps the layers separate:
| Decision question | Best starting source | What still needs validation | |---|---|---| | How large is the occupation and what does it pay? | OEWS | Current employer demand and local operating context | | Is the industry adding, losing, hiring or separating workers? | QWI | Which occupations are driving the movement | | What work, tasks and skills define the occupation? | O*NET | Local job design and employer-specific requirements |
The value comes from comparison, not forced agreement.
If OEWS shows a large occupation, QWI shows unstable industry employment, and employers report that the real constraint is a specialized skill not represented in the proposed curriculum, that disagreement is not noise. It is the decision.
What a reviewable conclusion looks like
Section titled “What a reviewable conclusion looks like”A useful workforce conclusion should state:
- which question each source was used to answer;
- the geography, industry and occupation definitions applied;
- the period covered by each dataset;
- where the sources agree;
- where they conflict or remain incomplete;
- which claims still require employer or provider validation;
- what evidence would change the recommendation.
This is more valuable than collapsing every signal into one score. A score can summarize evidence, but it should not erase the meaning or limitations of its inputs.
Start with the decision, not the dataset
Section titled “Start with the decision, not the dataset”Before opening a dashboard, write down the decision in one sentence.
Not: “We want to understand the semiconductor workforce.”
Better: “We need to decide whether to advance a Central Texas training concept for five semiconductor-relevant occupations to employer validation.”
That sentence determines the geography, occupation set, industry context, evidence threshold and next human review. It also reveals what public data cannot settle on its own.
Have a workforce question but are unsure which evidence belongs in the analysis? Bring us the decision, geography and sector. AvelinLabs can map the question to the right evidence layers before anyone builds a larger study.
Request evaluation access and bring the decision, geography, and sector you need to evaluate.
Sources
Section titled “Sources”- U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics: https://www.bls.gov/oes/
- U.S. Census Bureau, Quarterly Workforce Indicators: https://www.census.gov/data/developers/data-sets/qwi.html
- ONET Resource Center, ONET Database: https://www.onetcenter.org/database.html