Methodology

About this brief

The Approval Trends Brief publishes weekly snapshots of USCIS processing data covering 15 forms. Each issue summarizes the past 8 weeks of approval medians, breaks down by service center, and surfaces patterns the data reveals.

How we compute the numbers

Each issue leads with the total wait: the median number of days from the date USCIS received a case to its first approval, for cases approved in that ISO week. When fewer than 100 of that week's approvals, or fewer than half of them, have a known filing date, the issue says the total wait is unavailable and leads with the final-stage wait instead. When USCIS has not shown a case's received date, we estimate it from neighbouring receipt numbers; cases where neither is possible are left out of the figure.

The trend charts, stage table and RFE comparison measure a shorter span, the final-stage wait: days from a case's last status update before approval to the approval itself. It is not the total wait. Service center breakdowns group by the center that issued the approval, not the original receipt center.

Both figures are measured on approved cases, so neither describes how long pending cases have waited. USCIS's published processing time is a different statistic: the time within which 80% of cases in one category were completed, counted from filing.

The 8 week slope is a linear regression over the most recent 8 weekly medians. R² indicates how well that line fits; values above 0.7 suggest a strong trend, between 0.4 and 0.7 a moderate one, below 0.4 noise.

The 4 week forecast extrapolates the 8 week slope. It is a mechanical projection, not a prediction of policy or shocks.

Data sources

Source data: USCIS Case Status pages, scraped daily via our tracking infrastructure. Approximately 600,000+ active cases tracked across all forms.

About AI assisted sections

Some narrative sections are generated by an AI assistant and reviewed against rubric scores before publication.

Sections 4 and 9 of each issue, plus the headline and lede, are generated by an AI assistant using only the data context summarized above. The output is reviewed by an automated critic against eight quality dimensions before publication; drafts that fail are rewritten or fall back to template text.

Questions or corrections

Found a number that seems wrong, or a claim you would like to challenge? Email cs@mycaseshub.com.