How NIWStats collects and counts

Where each case comes from

Every figure on this site traces back to an individual EB2-NIW case, and every case carries one of four provenance labels. As of September 2026, the 907 cases break down as:

⟳ tracked (557) — receipt on file, still pending; our USCIS check reads the case status and re-checks it until the outcome lands.
✓ USCIS-confirmed (186) — receipt on file, decided; the outcome itself was read from USCIS.
owner-reported (23) — stated by the case owner and not yet USCIS-checked; receipt-backed cases move up after their first check.
compiled (141) — gathered from public community posts where applicants shared their outcomes.

743 of the 907 cases carry a receipt number we check against USCIS directly — on those, the outcome comes from USCIS, not from what anyone typed. Even so, only the status comes from USCIS; the field and profile are as the owner described them. Published figures are computed over all shared cases, and per-case provenance is always inspectable in the app. This is a crowdsourced dataset, not an official government release.

How current the data is

Our USCIS check re-reads every tracked case while it is pending: at least once a day, more often as a premium case nears the end of its 45-business-day clock, and hourly once it is inside the decision window. A decision, RFE, or other status change is recorded when that check first sees it — usually within the hour for premium cases in the window, and within a day for the rest — and appears on the app's decisions feed and queue as soon as it is recorded.

The figures on these pages are a snapshot of the same data, refreshed periodically; each page shows when it was last refreshed (this one: September 30, 2026). USCIS's own NIW counts are published quarterly; the approval-rate page quotes the latest quarter next to the tracked figures, labelled with its period.

What we ask

We collect the handful of things that genuinely shape how an NIW case goes: the broad endeavor field, the applicant profile (research, industry, or a mix), whether the case used premium processing, whether it drew an RFE, the rough timeline, whether it was filed with an attorney or on one's own — plus, optionally, a short endeavor line ("AI infrastructure", "battery materials") and which firm filed the case.

Most of these are broad categories. The two optional free-text fields stay short by design: the endeavor line is a phrase, not your proposed-endeavor narrative, and the firm name is used only in counts.

Composition and counts, not invented rates

We show composition and counts — how cases distribute across fields, profiles, RFE history, and timelines — and shares only with their n attached. We never publish an all-time approval rate over the whole dataset: our early compiled cases (the "seed") deliberately sought out the denials that usually go unreported online, so the seed over-represents denials relative to the real base rate — an approved-vs-denied percentage computed over it would mislead. That's also why blended approved and denied totals are never shown side by side. The one approval share we do publish is scoped to a slice the seed can't reach — decided under tracking, below — and always carries its n and date range.

Dated weekly movement counts ("3 approvals · 1 denial · 2 RFEs" for a completed week) are different: they're event counts of what our USCIS check observed that week, not a rate, and recent weeks don't carry the seed's skew. Weeks can fill in retroactively as new cases join with past-dated events.

The suppression thresholds are fixed and boring on purpose: a per-field share renders only when the cell has at least 8 cases with a known value (thinner cells show the raw count, faint); a field joins a ranked comparison only with at least 20 decided cases. Below a threshold you'll see a count, or nothing — never a confident-looking figure resting on almost no data.

"Decided under tracking", first actions, and the premium window

One scoped slice does show approved and denied counts together, and the approved share as a percentage: decided under tracking, on the approval-rate page. It counts only cases shared while still pending whose decision our USCIS check later read — outcomes added after the fact are excluded, because a decision already known when it's shared can be picked for how it turned out, and shared outcomes over-represent approvals. Within that slice, a group's outcome split renders only with at least 8 such decisions, and every interpretive sentence on that page is rebuilt from the counts at each refresh — it disappears if the pattern it describes does.

The RFE page's first-action mix uses the same scoping: only first actions — a straight decision, or an RFE — that landed on premium cases we were already tracking as pending. Every member of that pool has reached a first action, so no still-waiting case sits in any denominator.

The premium-processing turnaround figures share one definition everywhere they appear: the clock starts at the I-907 receipt date, counts business days (weekends and federal holidays excluded), and runs to USCIS's first action. Published figures follow the current pace — they cover the shortest recent window holding enough decisions, and every surface quoting them names that window. When an RFE response restarts the clock, the restarted run is counted separately, never against the 45-day promise. Business-day figures are approximate, since I-907 dates are shared by case owners.

What we never show

Two things stay private no matter what. The receipt number is used for exactly one job — reading the case's status from USCIS — and is never displayed anywhere, on any page, to anyone. Law-firm names never appear on an individual case: a case shows only whether it used an attorney, and a firm's name (where the owner chose to share it) will only ever appear in counts by endeavor field, once that firm has several cases. We also never ask for your name.

A data instrument, not legal advice

NIWStats exists to inform your own judgment with comparable cases. It is not, and cannot be, legal advice, and it is not affiliated with USCIS. For decisions about your own petition, talk to a qualified immigration attorney. The site stays free to use; it may show a clearly labeled sponsored link, always kept separate from the data.