Late summer is when priority dates go backward

In 24 of the 49 Augusts and Septembers on record, at least one employment-based priority date moved backward. In January through June: 26 of 149.

Based on 293 Visa Bulletins, December 2001 to September 2026.

80 months, one pattern

Each cell is one Visa Bulletin. Filled cells had at least one employment-based retrogression; darker means more categories moved at once. Tap any filled cell to open that bulletin.

How this is counted

• Conservative vs. sensational counting: Counting by individual row-months yields a multiplier of 4.6× (141/812 in August–September vs. 92/2458 in January–June). However, this inflates sample size because retrogression across multiple categories in the same bulletin stems from a single agency decision. Our primary headline multiplier of 2.8× is based on the conservative and statistically honest calendar-month unit.
• Era dependency: The retrogression risk multiplier varies across observation windows. In recent years (October 2015 – September 2026), the calendar-month ratio is 3.9× (and 7.4× in row-months), compared to 2.4× in the earlier period (January 2005 – September 2015). While the seasonal pattern holds across both eras, claiming an exaggerated ratio as a permanent rule would be inaccurate.
• Historical bulletin gaps: The dataset has 5 missing bulletin months (2009-03, 2009-09, 2009-10, 2009-11, 2012-10) due to gaps in public archives, out of 293 total calendar months.
• Legislative risk: A legislative change, such as the removal of per-country caps, could fundamentally alter the visa allocation mechanism and invalidate the historical pattern this model relies on. Retrogression reflects visa number exhaustion ahead of the September 30 fiscal year close, and future policy shifts can reshape these dynamics in ways the dataset cannot anticipate.

Sources and reproduction

• Raw dataset: /data/seasonal-risk.json
• Official bulletin archive: travel.state.gov