Introduction
A program viability memo has two halves. The supply half is yours: completions by CIP code, which you can count exactly. The demand half belongs to the labor market, and almost every federal number on it, wages, employment counts, projected openings, arrives sorted by SOC code. If you have ever pasted a BLS table into a memo without knowing what the code prefixes mean, or what "annual openings" actually counts, this is the primer.
What the SOC is
SOC stands for the Standard Occupational Classification. It is a federal statistical standard maintained by the Bureau of Labor Statistics, and its rule is deceptively simple: every worker is classified into exactly one occupation according to the work they perform, and in some cases the skills, education, or training needed for the work. The current edition, the 2018 SOC, classifies all workers into 867 detailed occupations, which are combined into 459 broad occupations, 98 minor groups, and 23 major groups. Where CIP describes what a program teaches, SOC describes what a job consists of. They are separate vocabularies maintained by separate agencies, and they do not overlap by design.
How a code is built
A SOC code reads like 15-1252: two digits, a hyphen, four digits. Each prefix narrows the group:
The four levels of a SOC code, using software developers
| Level | Code | Example |
|---|---|---|
| Major group | 15-0000 | Computer and mathematical occupations |
| Minor group | 15-1200 | Computer occupations |
| Broad occupation | 15-1250 | Software and web developers, programmers, and testers |
| Detailed occupation | 15-1252 | Software developers |
The residual codes
Every level ends in a catch-all: major groups hold "all other" minor groups, and detailed occupations ending in 999, like 15-1299 Computer Occupations, All Other, sweep up work that does not fit a named code. These residuals are usually where a fuzzy job title actually lands, so treat any analysis that leans on them with suspicion. If your local employer data shows heavy employment in an "all other" bucket, the honest reading is that the taxonomy cannot describe those jobs, not that a giant mystery occupation is growing.
Where you meet it
SOC is the spine under most labor market data the federal government publishes:
Federal data systems that run on SOC codes
| System | What it gives you | Cadence |
|---|---|---|
| OEWS, Occupational Employment and Wage Statistics | Employment and wage estimates by detailed occupation, nationally, by state, and by metro area | Published annually, spring release |
| Employment Projections | Ten-year projected employment change and annual openings by occupation | Published annually, late summer or fall |
| O*NET | Skills, knowledge, and task profiles attached to each detailed SOC code | Continuously updated |
| Occupational Outlook Handbook | The public-facing career profiles built from all of the above | Rolling updates |
Revisions, and why they bite
The SOC is revised roughly every eight to ten years; the 2018 edition replaced the 2010 edition through a Federal Register process that ran from the first notice in 2014 to final publication in 2017. Revisions split, merge, and add occupations, and the changes are not cosmetic. The 2018 SOC created a standalone Data Scientists occupation (15-2051) out of work previously buried in statisticians and survey researchers, and restructured software developers from the old 15-1132 into the new 15-1252. BLS publishes a 2010-to-2018 crosswalk, but the same caution as CIP applies: a wage or employment trend that crosses a SOC revision is two different taxonomies stitched together, and any jump or drop across the seam needs a footnote, not a narrative.
Reading SOC data as an IR office
- OEWS is an employer survey. It counts jobs where they are located, not graduates where they studied. Comparing your completions to OEWS employment in your service region is a market share argument, and should be labeled as one.
- Annual openings are mostly replacement needs, not new jobs. BLS projects openings from growth plus replacements from people leaving the occupation. A field with zero net growth can still show thousands of annual openings, and citing the openings number as if it meant growing demand is the most common error in program proposals.
- Granularity runs out at the state line. OEWS publishes state and metro estimates, but the ten-year national projections, including the openings figures, are national. There is no official state-level openings number; anyone who hands you one built it themselves, and their method deserves a look. We ran into exactly this when building our graduates-per-opening analysis.
- Wage numbers are for workers, not entry-level graduates. OEWS reports the occupational wage distribution across all experience levels. The median for an occupation is not a forecast of what a fresh completer earns; for that, the federal file that follows actual graduates is PSEO, not BLS.
Where to go next
The supply side of this story is CIP: see what is a CIP code for the program taxonomy. The bridge between the two, the federal crosswalk that maps programs to the occupations they prepare students for, is searchable by program or occupation in our CIP to SOC crosswalk tool. For a worked example of the whole pipeline, our graduates per opening analysis matches IPEDS completions against BLS projected openings for 17 occupational pairs. To browse the occupation side directly, every occupation page on Clema ranks fields of study by how many graduates per year they feed into an occupation.
Pull labor market data in plain English
Ask for wages, employment, or projected openings for any occupation and get the BLS series and vintage on every answer.
Book a demo