breakdown of revisions by month.

The benchmark process replaces March data with the QCEW data and then wedges the difference back through the previous April.

You can compare the QCEW data for each month to the CES data for that month but BLS doesn't replace the CES data with QCEW data for any month other than March and in fact would caution that this comparison would be misleading. And that doesn't even factor in the impact of seasonal adjustment. QCEW data is not seasonally adjusted, CES is.

But I did the comparison anyway and 90% of QCEW/CES difference from March was already in place in January 2025. I will note, however that the gap between the CES & QCEW data spiked significantly in December 2024 and continued to grow through March.

Note that last year, the initial estimated revision for March 2024 was 818,000, but the final revision was just 589,000.

Also note that the divergence of the 2 programs spiking in December 2024 and and growing each month after that makes some sense. Since Trump was elected in November, it should not be a surprise that job growth could sharply decline immediately following that, particularly with his promise of job killing tariffs.