Every instrument we run rests on one thesis, so it should be stated plainly and then tested against the data: the labor market is switching units. For a century the unit was the title, a guild word that bundled skills, status, and pay into one string. The bundle is coming apart. Postings now describe work as skill lists; employers state outright which adjacent backgrounds they welcome; the same skill set gets repriced under three different names. The title is becoming what the filename is to the file: still useful, no longer the thing itself.
That number is our own pipeline showing the seam. We read 182,773 raw postings in the July run (100,215 after cross-board dedup), and the single hardest engineering problem is not extracting skills, it is surviving the titles: the same job arrives as five spellings, and tens of thousands of title strings are pure invention. The skills underneath, by contrast, cluster hard enough to measure. That asymmetry, chaotic labels over stable skill demand, is the whole case in one dataset.
Skills are specific. That is the discipline.
The skills-based market is routinely oversold as “your skills can take you anywhere.” The data says nearly the opposite: across the 3,521 occupation-to-occupation pairs we score, 55 percent sit under 20 percent readiness, and only 60 pairs anywhere in the matrix reach 60 percent. Skills are not a universal solvent; they are a specific inventory with a specific reach. That specificity is exactly why they carry value, and why measuring the reach beats asserting it.
Skills are specific. That is why they are worth money.
The careers whose skills travel
Count each occupation’s routes out at 45 percent readiness or better and the market sorts itself. The wide-exit careers share a shape: either a data core (analysis, SQL, statistics travel almost everywhere) or a coordination core (the operations-and-stakeholders bundle that every industry buys).
| Career | Routes out | The portable core |
|---|---|---|
| Sales engineer | 13 | Technical depth + commercial motion |
| Operations manager | 13 | Coordination, process, P&L adjacency |
| Data scientist | 12 | The full data stack |
| Account executive | 9 | Pipeline, negotiation, CRM |
| Business analyst | 9 | Requirements, analysis, process |
| Psychologist | 7 | Assessment, research, casework |
| Data analyst | 6 | SQL, visualization, statistics |
| Medical assistant | 6 | Clinical floor skills, patient ops |
At the other end, a set of large occupations show zeroroutes out at 45 percent in our matrix: copywriter, creative director, social media manager, executive assistant, recruiter, translator, paralegal, UX writer. Two honest readings, and both matter. First, craft careers run on deep, narrow vocabularies the wider market does not post for by name, so their exits really are harder at full readiness. Second, our instrument reads posted skill demand, and postings under-specify craft depth; a copywriter’s judgment shows up in postings as a thin skill list. The lock is real, and it is also partly a measurement shadow. We say so rather than pretend the number is complete.
The careers most open to outsiders
Reverse the lens and count how many origins reach each destination at 45 percent or better. The most skill-open doors in the market right now, measured: solutions architect and conversation designer (8 origins each), then database administrator, prompt engineer, executive assistant, and customer success manager (6 each). Notice what tops the list: the AI-era titles. They are young enough to have no guild, no credential wall, and no settled pedigree, so they hire on demonstrated skills because there is nothing else to hire on. New titles are the skills-based market in its purest form.
Notice also the asymmetry hiding in the two lists: executive assistant and customer success manager are among the easiest doors in (6 origins each) and among the hardest doors out (zero routes at 45 percent). A career can be skill-friendly to enter and skill-locked to leave. If you are choosing a landing spot for a pivot, that difference is worth more than the title on the door.
The passports: skills that cross the most borders
If the market’s unit is the skill, the highest-leverage question becomes: which skills appear in the most occupations’ demand? From the top-20 posted-demand profiles of all 177 occupations:
| Skill | Occupations |
|---|---|
| Data analysis | 62 |
| Supply chain | 47 |
| Training & facilitation | 46 |
| Project management | 44 |
| Customer service | 42 |
| Lean / Six Sigma | 41 |
| LangChain / agents | 33 |
| LLMs / generative AI | 31 |
| Python | 29 |
| SQL | 26 |
Two things stand out. Data analysis is the market’s reserve currency, demanded in a third of everything. And the 2026 story is already in the table: agent and LLM tooling sits in the top-20 demand of roughly a third of occupations, two years after those phrases barely existed in postings. A bridge skill is leverage precisely because one investment moves your readiness across dozens of destinations at once; the instrument prices that for your specific starting point.
The title is the label. The skill set is the asset.
What to do with the thesis
Three moves follow from the data, none of them motivational. Choose destinations by measured overlap, not title glamour: a 60 percent route you can close in months beats a 15 percent route with a better-sounding name. Learn bridge skills before niche skills when you are undecided; data analysis and the LLM toolchain buy readiness across whole regions of the market. And respect the second axis: where a license stands, no skill overlap shortens it, which is why every gated route on this site says so in plain text.
All figures are computed from the PivotHop July 2026 corpus: 182,773 raw postings, 100,215 after cross-board dedup, mapped onto 177 occupations; 42,254 distinct unmapped title strings from the same run’s pipeline counters. Route counts use the full adjacency matrix (3,521 scored directed pairs; pairs sharing too few skills are unscored); “routes out/in” count pairs at ≥45% readiness; bridge-skill counts are occupations whose top-20 posted-demand profile (≥3 postings and ≥2% share) includes the skill. Readiness methodology, including its limits, is in What is career adjacency; the craft-vocabulary measurement caveat above is a real limit of posted-demand data and we state it rather than smooth it.