ADP Research explains how its January employment report is built

ADP Research explains how its January employment report is built

ADP Research published this account on February 10, 2026, explaining how the January National Employment Report (NER) is constructed and revised. ADP says its NER uses payroll data from employers that choose its services, then weights the data using employment characteristics measured by the Quarterly Census of Employment and Wages (QCEW), a Bureau of Labor Statistics measure. ADP says the QCEW arrives with a lag, so the NER aims to provide a timelier measure of U.S. private-sector employment. ADP says its January release includes annual benchmarking to QCEW data through March 2025 and, beginning with this release, more frequent adjustments for bias in its payroll sample. The source says that sample tends to include more resilient and faster-growing employers than the full population. It also describes weekly seasonality estimates, revisions to historical data, and the relationship between the NER and NER Pulse. ADP cautions that pay-frequency differences make recent weekly estimates more variable, so it withholds recent data in its publications. The account ends by stressing that job creation is only part of a healthy labor market, with wages another critical component; ADP says it will next discuss its payroll data on pay insights. How well the adjustments capture employment changes between QCEW releases remains unknown from this account.

Publication: 87fba424-21ae-fa4c-1d1c-43a5c94c8b0e

attribution — pass: This is a single ADP Research Main Street Macro piece. The reading attributes its statements about the NER, QCEW benchmarking, bias adjustment and the NER Pulse to ADP throughout, and does not present them as established fact.

Here are five things you should know about the NER January release and our benchmarking process.

summary — uncertain: The summary accurately reflects the five points: annual benchmarking, quarterly bias adjustment, weekly seasonality, coarser historical granularity and the NER/Pulse relationship. Two issues affect reader understanding. First, the only page definition for benchmark describes AI test suites, but here benchmarking means reweighting payroll data to QCEW census counts, so the tooltip misleads. Second, the reading never explains this methodology note's relevance to AI; the source mentions AI only in passing. Smallest repair: add a statistical-benchmarking term entry and note the limited AI link.

As the labor market adapts to artificial intelligence, demographic change, and business cycle fluctuations

date — pass: The originalPublished day, February 10, 2026, is consistent with the text: the January NER was released the prior week, and the week-ahead previews the delayed BLS January report. No hindsight framing appears. The evidence kind is publisher-page, so no archive disclosure is required, and the retrieval date is noted.

Last week we published the first ADP National Employment Report for 2026.

support — pass: Each claim has a matching verbatim excerpt. The summary's statements about bias, the QCEW lag, withholding recent weeks and wages are source-stated. The closing question and unknown are framed as open rather than as conclusions, and no benefits are implied.

By adjusting the growth bias in the ADP payroll sample more frequently, we can better capture employment changes in real time.

media — pass: No media is selected and no verified candidates were supplied, so nothing is misidentified.