Final Refusals Fell Every Month of Q2 Even as Examiners Issued More Non-Finals

Quarterly USPTO Prosecution Report | USPTO data through June 30, 2026

Something split apart inside the USPTO's examining corps this spring. Non-final office actions climbed to their quarterly high in June — 38,166, up 16.2% from May — while final actions did the opposite, falling in each of the three months from 4,366 in April to 3,383 in June. The quarter's busiest month for first-round refusals was also its quietest for final ones. The rebound everyone will notice in the headline volume number was built entirely on non-final actions.

That divergence is the story of the quarter, and it is worth more than the month-over-month wiggle that usually leads these reports.

When the volume rebounds but the final refusals keep falling, the pipeline is filling faster than it's closing — and every docketing team should be staffing for the front end, not the back.

— Howard Katzenberg, Founder, GleanMark

We think this most likely reflects timing rather than a change in examiner posture: a surge of newly examined applications hitting first action in June, with the corresponding finals still months downstream. If that read is right, the final-action count is the one to watch next quarter.

Executive Summary

• The split: Non-final actions rose to a quarterly high of 38,166 in June, but final actions fell every month — 4,366, then 3,644, then 3,383. The June rebound in total office-action volume came entirely from non-finals.
• Headline volume: The USPTO issued 105,832 non-final and 11,393 final trademark office actions in Q2 2026, for 117,225 office actions total.
• Section 2(d) dominates the analyzed refusals: Likelihood of confusion was the primary refusal in 11,128 of 23,205 analyzed responses — 48.0%, more than the next four categories combined.
• Best-performing argument: Disclaimer arguments led all categories on directional effectiveness at 0.840, ahead of class deletion at 0.802 — but read that alongside the legal issue it actually solves, not as a general win rate.

1. Prosecution Volume

Period	Non-Final Actions	Final Actions	Total Office Actions	Non-Final MoM Change
April 2026	34,811	4,366	39,177	—
May 2026	32,855	3,644	36,499	-5.6%
June 2026	38,166	3,383	41,549	16.2%

Quarter	Non-Final Actions	Final Actions	Total Office Actions	All Prosecution Documents
Q2 2026	105,832	11,393	117,225	541,526

These are total USPTO figures, not the narrower analyzed subset used for the refusal and argument analysis further down.¹

The non-final path was not a climb — it was a dip and a bounce. Volume fell from April to May, then jumped past both prior months in June. Total office actions traced the same V.

Final actions did not. They fell in a straight line across all three months. So the two curves that usually move together came apart: the busiest month for first refusals was the slowest for finals. Read the June number as a front-loading of the examination pipeline, and expect the finals it implies to land in Q3 and Q4.

2. Office Action Document Types

Document Code	Document Type	Documents in Quarter
NFIN	Non-Final Action	105,832
NFO	Notice of Non-Final Office Action	102,311
ROA	Response to Office Action	62,931
ABN	ABANDONMENT-Notice	38,316
AMC	Amendment and Mail Process Complete	29,619
PB1	Notice of Publication	15,470
EB4	Notification of Notice of Publication	15,470
PST	TRAM Snapshot of App at Pub for Oppostn	15,208
PB3	OG Publication Confirmation	14,418
ORC	Registration Certificate	12,320
ERO	Notice of Registration Email	12,073
FREF	Final Action	11,393
XSS	XSearch Search Summary	9,260
ENA	Notice of Allowance	8,823
SPE	Specimen	6,699
PNR	Public Note	6,490

One trap to flag: NFIN (Non-Final Action) and NFO (Notice of Non-Final Office Action) are separate codes. Do not add them together when you measure office-action volume — that double-counts nearly every first refusal in the quarter. The rest of the table is a reminder of how much prosecution work happens away from examiner refusals: 62,931 responses, 38,316 abandonment notices, and 12,320 registration certificates all cleared in the same three months.

3. Refusal Landscape

Primary Refusal Type	Analyzed Responses	Share of Analyzed Responses	Registered to Date
Section 2(d) — Likelihood of Confusion	11,128	48.0%	4,859
Section 2(e) — Descriptiveness / Surname	4,602	19.8%	2,167
Identification of Goods/Services	2,741	11.8%	1,382
Specimen Refusal	2,615	11.3%	985
Other	1,838	7.9%	1,101
Disclaimer Requirement	166	0.7%	97
Ornamental Refusal	95	0.4%	39
Section 2(c) — Name / Portrait / Signature	9	<0.1%	6
Section 2(a) — Deceptive / False Association	7	<0.1%	2
Failure to Function	4	<0.1%	0

Refusal Type (Overlapping Tag View)	Tagged Analyzed Cases	Tagged Cases Registered to Date
Section 2(d) — Likelihood of Confusion	11,128	4,859
Identification of Goods/Services	8,573	3,869
Section 2(e) — Descriptiveness / Surname	8,005	3,617
Disclaimer Requirement	6,580	3,282
Specimen Refusal	5,058	1,969
Other	1,933	1,140
Ornamental Refusal	305	125
Section 2(a) — Deceptive / False Association	49	15
Failure to Function	30	10
Section 2(c) — Name / Portrait / Signature	16	9

Likelihood of confusion is not just the largest refusal — at 48.0% of analyzed responses, it is bigger than descriptiveness, identification, specimen, and everything in "other" put together. Section 2(e) is a distant second at 19.8%.

The two tables tell different stories, and the gap between them is the useful part. Identification issues are the primary refusal in 2,741 cases but show up as a tag in 8,573 — more than three times as often. Disclaimer requirements are primary in just 166 cases but tagged in 6,580, roughly forty times as often. Read plainly: identification and disclaimer problems rarely stop an application on their own. They ride along with the substantive refusal, and a clean amendment clears them while the real fight is about registrability.

Because the tag counts overlap, they cannot be summed into a quarter-wide refusal total or converted into a share of all USPTO office actions. The primary table is the clean percentage view; the tag table is the co-occurrence view.

4. Argument Effectiveness

Argument Category	Directional Score (0–1)	Argument Instances
Disclaimer	0.840	3,296
Class Deletion	0.802	650
Consent Agreement	0.757	467
Supplemental Register	0.739	1,013
Secondary Meaning	0.731	799
Incongruity	0.719	119
Commercial Impression	0.715	1,370
Goods Coexistence	0.713	1,650
Cited Mark Abandoned	0.709	631
Weakness of Cited Mark	0.700	823
Third Party Registrations	0.697	467
Suggestive Not Descriptive	0.695	1,270
Market Channel Distinction	0.692	839
Double Entendre	0.677	50
Specimen Support	0.672	1,780

Refusal Type	Directional Score (0–1)	Argument Instances	Verified-Win Instances
Section 2(d) — Likelihood of Confusion	0.696	11,629	2,429
Disclaimer Requirement	0.778	4,027	2,076
Specimen Refusal	0.714	2,302	700
Identification of Goods/Services	0.679	1,848	628
Section 2(e)(1) — Descriptiveness	0.753	1,278	371
Other	0.665	536	249
Section 2(e)(2) — Geographic Descriptiveness	0.762	137	46
Ornamental Refusal	0.717	114	54
Section 2(e)(4) — Surname	0.728	100	36
Section 2(a) — Deceptive / False Association	0.697	25	5
Section 2(e)(5) — Functionality	0.759	14	1
Section 2(e)(3) — Geographic Deceptiveness	0.868	12	6
Failure to Function	0.665	2	0

The unit here is argument instances, not cases — one response can raise several. The directional score is a 0–1 signal, not a win rate.

Disclaimer tops the category table at 0.840 across 3,296 instances, with class deletion (0.802) and consent agreements (0.757) behind it. But the score only means something once you match the argument to the refusal it answers. Disclaimers clear descriptiveness or genericness of a component of the mark; they do nothing for likelihood of confusion. Consent and coexistence agreements are the Section 2(d) tools. Supplemental Register is a Section 2(e) play, and only for use-based applications. A high score on the wrong argument is just a high score.

By refusal type, Section 2(d) generates by far the most argument volume — 11,629 instances at a 0.696 score. The eye-catching 0.868 belongs to Section 2(e)(3), geographic deceptiveness, on a grand total of 12 instances. That is a headline waiting to mislead someone. Weight it accordingly.

Practitioner Takeaways

1. Staff the front of the pipeline. June was the quarter's high for non-final actions while finals kept falling. If that gap is a timing effect, the response and final-action wave lands next quarter — plan capacity there now.
2. Section 2(d) is the whole game. It was the primary refusal in nearly half of analyzed responses. Goods narrowing, market-channel distinctions, and consent agreements are the toolkit; disclaimers and Supplemental Register are not.
3. Clear the co-occurring requirements early. Identification and disclaimer issues show up far more often as tags than as primary refusals. A clean amendment removes the procedural drag while you argue the substance.
4. Use directional scores for triage, not prediction. A 0–1 score is a signal, not a probability. Discount high scores built on thin samples, especially in narrow refusal types.

Explore more USPTO data analysis on the GleanMark Insights blog.

Footnotes

1. Refusal and argument findings come from GleanMark's analyzed subset — 23,205 analyzed Q2 responses and 12,432 analyzed office actions — against 105,832 total USPTO non-final actions in the quarter. It is not a census of all USPTO refusals. Data sourced from USPTO TSDR and public prosecution records, as recorded through the report date. This report is informational and does not constitute legal advice. ↩