The Count Rush

§ 1 · The sprint

Four peaks a year, every year, on the fiscal calendar.

USPTO examining attorneys work against production goals measured on the agency's fiscal calendar — the year ends September 30, and the quarters close with December, March, June, and September. Output follows the goalposts. Across 2022 through mid-2026, first-action volume in quarter-closing months ran 21–32% above the other months of the same year — every year, all four quarters. The drop on the far side is just as sharp: March 2023 produced 33,611 first actions; April 2023 produced 20,911, a 38% fall across a single month boundary. March — the mid-year checkpoint — is the tallest peak in every year of the window.

None of this is scandalous — deadline-shaped output is how production systems behave everywhere. What matters for the applicant is that the sprint doesn't just change how much gets done. It changes what gets done.

§ 2 · The mix

The close approves. The month after refuses.

For every application there is a first look — the moment the examiner either approves the mark for publication on sight or opens with an office action. Pooled across the window and restricted to represented domestic filings (the composition control; pro-se and Madrid filings are tracked separately and tell the same story), the share of first looks that open with an office action is not flat across the year:

§ 3 · The Final that waits for October

September is the year's low for Final refusals. October is the high. Every year.

The same calendar shows up on applications already in a refusal fight — and this is the part selection can't explain away, because these are decisions on the same file. When an examiner resolves a pending refusal (issues a Final, or approves the application after the office action), the split depends on the month: of fights resolved in the pooled Septembers, 17.3% ended in a Final refusal. In the pooled Octobers, 20.1%. Writing a Final at the close is the count-poor move — it creates future work instead of booking a disposal — so Finals are systematically deferred past the fiscal-year end. In 2024 the fleet issued 3,471 Finals in September and 5,501 in October.

§ 4 · Five ways this could be nothing

Four ruled out. One partly true — and priced in.

1. Hypothesis 01

   “It's the applicant mix, not the examiner.”

   Ruled out

   The swing survives inside narrow strata. Within represented domestic filings alone, July runs 6.3 points above June; within pro-se domestic filings, 4.6 points. The pro-se share of first looks barely moves month to month (23–25%), and Madrid §66(a) filings — which essentially always open with an office action — are excluded from the controlled series entirely.

2. Hypothesis 02

   “It's calendar mechanics — long months, short months.”

   Ruled out

   Business-day counts vary by roughly 5% month to month; the output spikes run 21–38%. March out-produces both February and April by about a third with comparable working days.

3. Hypothesis 03

   “It's the pandemic backlog era.”

   Ruled out

   The window starts in 2022, and excluding the backlog signature — first actions landing more than two years after filing — moves the monthly rates by less than half a point. The pattern repeats cleanly in 2024, 2025, and the first half of 2026, and July sits above June in all four complete years.

4. Hypothesis 04

   “It's a data artifact.”

   Ruled out

   Every date here is the USPTO's own: action dates are the examiner-signed dates in the agency's daily XML, and approvals are the date of the examiner's approved-for-publication event. Nothing is timestamped by our pipeline.

5. Hypothesis 05

   “Examiners pick different files, not different answers.”

   Partly true — and it doesn't change the read

   This one deserves honesty: some of the swing isselection. An examiner hunting counts at the close reaches for the applications that can be approved without writing anything — your mark's merits don't change, but when it gets picked up does. Two things say selection isn't the whole story. The swing holds inside narrow, homogeneous strata, and — more telling — decisions on already-pending refusal fights move with the same calendar (§ 3): those are choices on the same file, not choices among files. Either way, the practical effect on your docket is identical.

§ 5 · What this means for your prosecution

One lever, one non-lever, one reading rule.

1. 01

   Don't re-time your filings over this.

   The queue from filing to first look runs six to nine months with real variance — you cannot reliably aim a filing date at a friendly examination month. If the filing calendar matters to your client for other reasons, let those reasons win.

2. 02

   Your response date is the lever you actually hold.

   Examiners typically act on a response within a month or two. A response filed in July or August tends to be decided into the September fiscal-year close — the month where refusals are most likely to come off and least likely to go Final. A response filed just after a quarter opens lands in the toughest stretch of the cycle. When the six-month response window gives you room, land the decision on a quarter close, not right after one.

3. 03

   Read an early-quarter office action in context.

   A first action dated July or October is the seasonal norm, not necessarily a signal about your examiner. Whether the examiner actually runs tough — and what overcomes them — is a per-examiner question; that's what the profile is for.

Methodology