What the numbers mean
FillingTide publishes the USCIS processing time as one value: the time it took USCIS to complete 80% of adjudicated cases over the prior six months. It is a backward-looking operational statistic, not a deadline, guarantee, or personalized estimate.
The separate case-inquiry threshold reflects the 93rd percentile and helps USCIS calculate when a case may be outside normal processing time. We never combine the 80th- and 93rd-percentile values into a range. Both source values retain their original units and are also normalized to months for validation; weeks use 52 weeks per 12 months and days use 365.2425 days per 12 months.
Collection and publication
- Retrieve a dated SQLite release from the independently maintained USCIS data mirror described on our data-sources page.
- Select only the launch combinations explicitly allowed in the pipeline configuration.
- Validate identifiers, required fields, units, unique page keys, history depth, and page eligibility.
- Write immutable dated snapshots and one minimal published dataset.
- Build static HTML from the validated dataset. The public site does not query USCIS during a visit.
Trend calculation
“Faster” and “slower” compare only the 80th-percentile processing time in the latest release with the earliest snapshot available on that page. A lower value is labeled faster, a higher value slower, and an equal value unchanged. The 93rd-percentile case-inquiry threshold is shown for reference but is never used as the trend series.
The 30- and 60-day views use a stricter rule: they select the nearest earlier publication with the same methodology only when its real interval is within 15 days of the target. The page states that actual interval. If no observation qualifies, the comparison is unavailable; FillingTide does not interpolate a value.
We do not interpolate missing observations, smooth the series, or turn the history into a forecast. A change can reflect workload, case mix, policy, measurement changes, or other factors that this dataset cannot establish.
Page eligibility and corrections
A detail page is published only when its latest 80th-percentile value is present and at least two real snapshots exist. Validation also checks source-asset hashes, publication dates, units, percentile ordering, and stable category and office meanings. Large changes must match an explicit human-review record or the update fails. Directory pages must contain meaningful child records; empty parameter combinations are not generated and return a 404.
If you find a source mismatch, send the page URL, the value in question, and supporting evidence to corrections@fillingtide.com. We will correct the published dataset rather than silently rewriting old source snapshots.
Method last reviewed: August 6, 2026.