Published October 11, 2026; sources checked October 7, 2026.
The decision problem
Job-to-job flows separate direct employer changes from other labor-market movement. That distinction matters when a recruiting team sees many accepted offers but cannot tell whether it competes mainly with other employers, labor-force exits, or local growth.
Build the extract
Use the Census Job-to-Job Flows data to compare hires, separations, job stayers, and origin-to-destination flows by geography and industry. A direct job change is evidence of movement between employers, not proof that pay, a manager, or a recruiting message caused the move. Start with counts, then calculate rates only with the published denominator.
Read the measure in context
Build a quarterly table with hires from employment, hires following nonemployment, separations to another employer, separations to nonemployment, and continuing jobs. Keep origin and destination characteristics straight. When only destination characteristics are selected, the extraction describes hires. Origin characteristics describe separations. Selecting both supports flow comparisons.
What an employer can do with the result
A recruiting team can use the pattern to decide where deeper local work is justified. A high share of job-to-job hires may support a competitor and occupation review. A high share arriving from nonemployment points to a different intake question: whether screening, start timing, transport, credentials, or schedule requirements keep interested workers from entering. Neither pattern sets an automatic sourcing strategy.
Limits that belong beside the finding
Do not turn an aggregate flow into a claim about a named competitor or employee. State suppression, industry coding, geography, revision, and time-window limits. Compare the same definitions over several quarters and mark shocks, boundary changes, and known coverage gaps.
Study controls and decision rules
Write the question and the smallest decision it could affect before collecting results. Name the population, geography, industry definition, period, comparison, exclusions, and data vintage. Record every transformation from source file to table. Another analyst should be able to reproduce the headline values without relying on private notes.
Predefine three outcomes: a stable pattern worth investigating, no material pattern under the chosen definition, or evidence too incomplete to interpret. None of those outcomes establishes causation. An aggregate difference can narrow the next question, but it cannot explain an individual's conduct or support an employment decision.
Check sensitivity by changing one defensible choice at a time. Test adjacent periods, alternate geographic boundaries, and broader industry groups. Mark suppressed and missing cells rather than converting them to zero. If the conclusion changes with a reasonable definition, describe it as unstable.
Privacy and fair-use boundaries
Use public workforce data to improve a process, not to label people. Do not combine small public cells with internal records in a way that could reveal an employee. Restrict internal extracts by role, keep only fields needed for the stated question, log corrections, and set a deletion date.
Aggregate findings must not feed discipline, scheduling, selection, pay, or promotion decisions about a person. Employment practices still need job-related evidence and consistent administration. Qualified owners should review legal, privacy, labor, safety, and equal-employment questions within their authority.
Reporting checklist
The final brief should include the source title and URL, access date, release or vintage, exact variables, filters, units, numerator and denominator where relevant, missingness, revisions, and unresolved questions. Show a table of values before presenting a chart. Label analysis and inference separately from source facts.
Keep the result modest. A credible brief often ends with one bounded follow-up, such as checking an internal queue, auditing a definition, or testing a process correction. It should never end with a fabricated benchmark, a universal target, or a promise that one intervention will reduce turnover.
Conclusion
Federal workforce data is most useful when the question is narrow and the limits remain visible. Preserve the source definitions, reproduce the result, and use the finding to choose the next investigation. Stop if the data cannot support that choice.
Reproducibility audit
Check 1: Origin and destination industry codes
Reconcile origin and destination industry codes across the origin, destination, and combined-flow extracts. Save the selected geography and quarter with the table. A disagreement requires a definition check before interpretation. State what the flow records and what it cannot reveal about a worker decision.
Check 2: Moves that cross county boundaries
For moves that cross county boundaries, compare counts before calculating a share. Inspect neighboring quarters and note revisions. Keep employer changes separate from entries out of nonemployment. Do not attach an explanation to the observed movement unless another suitable source supports it.
Check 3: Hires following nonemployment
Trace hires following nonemployment to the Census variable and extraction setting. Have a second reviewer reproduce the value. Describe suppression or missingness in the result, then decide whether the remaining evidence can answer the original recruiting-demand question.
Check 4: Separations followed by nonemployment
Reconcile separations followed by nonemployment across the origin, destination, and combined-flow extracts. Save the selected geography and quarter with the table. A disagreement requires a definition check before interpretation. State what the flow records and what it cannot reveal about a worker decision.
Check 5: Continuing jobs in the denominator
For continuing jobs in the denominator, compare counts before calculating a share. Inspect neighboring quarters and note revisions. Keep employer changes separate from entries out of nonemployment. Do not attach an explanation to the observed movement unless another suitable source supports it.
Check 6: Quarter-to-quarter revisions
Trace quarter-to-quarter revisions to the Census variable and extraction setting. Have a second reviewer reproduce the value. Describe suppression or missingness in the result, then decide whether the remaining evidence can answer the original recruiting-demand question.
Check 7: Flows within the same industry
Reconcile flows within the same industry across the origin, destination, and combined-flow extracts. Save the selected geography and quarter with the table. A disagreement requires a definition check before interpretation. State what the flow records and what it cannot reveal about a worker decision.
Check 8: Flows into a different industry
For flows into a different industry, compare counts before calculating a share. Inspect neighboring quarters and note revisions. Keep employer changes separate from entries out of nonemployment. Do not attach an explanation to the observed movement unless another suitable source supports it.
Check 9: Suppressed origin-destination cells
Trace suppressed origin-destination cells to the Census variable and extraction setting. Have a second reviewer reproduce the value. Describe suppression or missingness in the result, then decide whether the remaining evidence can answer the original recruiting-demand question.
Check 10: The distinction between a transition and a motive
Reconcile the distinction between a transition and a motive across the origin, destination, and combined-flow extracts. Save the selected geography and quarter with the table. A disagreement requires a definition check before interpretation. State what the flow records and what it cannot reveal about a worker decision.
Flow-direction cross-check
Build a small matrix with origin industries as rows and destination industries as columns. Reconcile row totals to separations and column totals to hires under the same filters. Mark within-industry moves on the diagonal and resist calling them competitor losses without employer-level evidence. Review whether a large cell persists across adjacent quarters or appears only once. Record the extraction date because later releases can revise values. If suppression prevents reconciliation, narrow the claim instead of estimating the missing flow. This cross-check tests the table, not the motives of workers represented in it.
Data Sources and Methodology
This guide reviews 10 public sources relevant to workforce data methods. The sources were selected for definitional clarity, public availability, and direct relevance to the operating question. Figures retain the reporting period and definitions used by the cited publisher. The article separates national or sector context from company-specific inference, labels any planning example as a hypothesis, and does not present an association as proof of causation. Source titles and retrieval links appear at the end of the guide. Local policy, contracts, and qualified professional review still govern decisions.
Sources
- Job-to-Job Flows Explorer, U.S. Census Bureau. Checked October 7, 2026.
- LED Extraction Tool, U.S. Census Bureau. Checked October 7, 2026.
- Job-to-Job Flows Data, U.S. Census Bureau. Checked October 7, 2026.
- Business and Workforce Dynamics Datasets, U.S. Census Bureau. Checked October 7, 2026.
- JOLTS Handbook of Methods, U.S. Bureau of Labor Statistics. Checked October 7, 2026.
- Statistical Quality Standards, U.S. Census Bureau. Checked October 7, 2026.
- Privacy Framework, NIST. Checked October 7, 2026.
- Employment Tests and Selection Procedures, EEOC. Checked October 7, 2026.
- Worker Well-Being Questionnaire, NIOSH. Checked October 7, 2026.
- FLSA Recordkeeping Requirements, U.S. Department of Labor. Checked October 7, 2026.
For help connecting workforce evidence to a repeatable recruiting operation, review StopHighTurnover services or contact the team.