Refusals 101, Part 1: What an Office Action Actually Is — and How Examiners Decide

Most people assume a trademark examiner sizes up your application and gives a thumbs up or down based on judgment. That's not how it works. When an examiner refuses a mark, they write it up the way a judge writes an opinion — citing a manual, citing prior decisions, applying a named legal test with a paper trail. The refusal you get is closer to a mini-brief than a verdict.

That distinction matters the moment your application hits a snag, because it tells you what you're actually fighting. Some objections are clerical. Others are legal battles decided by fifty years of precedent. Knowing which is which — before you name your company — is the difference between a two-week fix and a dead brand.

This is Part 1 of a series walking through why trademarks get refused. Start here, because everything else in the series assumes you know what an office action is and how the examiner behind it thinks.

The examiner's letter, and the clock it starts

An office action is the formal letter a USPTO examining attorney sends when something in your application needs to change or be defended. It lists every problem the examiner found — from a typo in your goods description to a full-blown refusal — and it starts a deadline.

There are two flavors. A non-final office action is the examiner's first pass: here are the issues, here's your chance to respond. A final office action comes after you've responded and the examiner still isn't satisfied on at least one point — "final" meaning your next move is either a narrow last argument or an appeal, not another round of back-and-forth.

Getting one is not a sign you did something wrong. Office actions are a normal part of examination.

The response clock is the part founders underestimate. For most applications you have three months from the date the office action issued, and you can buy one three-month extension for $125; applications filed through the Madrid Protocol get six months. Miss the deadline, or leave an issue unanswered, and you risk losing the application. The USPTO sometimes allows extra time to fix an honest omission in an otherwise complete response, but never count on it. The USPTO has a standard notice of abandonment for exactly that outcome. An incomplete or missing response isn't a soft warning. It's how names die quietly.

Examiners don't decide on vibes — they decide on precedent

Here's the part that surprises newcomers and reassures anyone who's ever felt an examiner was being arbitrary. Examiners work inside a strict hierarchy of authority.

At the base sits the TMEP — the Trademark Manual of Examining Procedure, the USPTO's own rulebook that tells examiners how to handle every recurring situation. Above the manual sit precedential TTAB decisions — rulings from the Trademark Trial and Appeal Board, the USPTO's in-house court that hears trademark disputes and appeals. Above the Board sits the Federal Circuit, the appeals court whose trademark rulings bind the Board and examiners. Above all of these sit the trademark statute itself (the Lanham Act), the USPTO's rules and, at the top, the Supreme Court.

An examiner refusing your mark cites these sources the way a judge cites case law. When the refusal names In re du Pont, that's not decoration — it's the examiner showing their work under authority they're required to follow. Which means your response has to engage the same authority. You don't argue with an examiner's opinion. You argue with their precedent.

What actually fills office actions

Strip an office action down to its component objections and many of them turn out to be fixable paperwork, not existential refusal.

Identification requirements — the examiner telling you to fix or clarify your list of goods and services — are among the most routine issues in examination. Specimen requirements — problems with the proof-of-use you submitted — are common too. Disclaimer requirements — the examiner asking you to formally give up exclusive rights to a generic or descriptive word inside your mark — are common for marks that pair a brand name with ordinary words describing the product.

None of those is a rejection of your brand. They're requests to tidy the application. A vague goods description gets rewritten. A bad specimen gets swapped. A disclaimer gets added. Annoying, deadline-bound, but rarely fatal.

The substantive refusals are the real fights, and two of them do most of the work. Section 2(d) confusion refusals — the examiner concluding your mark is too close to an existing one — are among the most common substantive grounds. Section 2(e)(1) descriptiveness refusals — the examiner saying your mark just describes what you sell, so it can't be registered on the Principal Register unless you prove buyers already see it as your brand — are the other big one.

The teaching point for a founder scanning an office action: find the substantive refusal first. Paperwork you fix. A 2(d) or 2(e) refusal you litigate, in miniature, against precedent. Those are the ones that decide whether you keep the name.

The cases in every examiner's arsenal

Because examiners cite cases, a handful of decisions do enormous work across the register. These are the names you'll see over and over in confusion and descriptiveness refusals — and the ones a junior practitioner should be able to recognize on sight.

One naming note before the list: in refusal letters the foundational confusion case shows up cited several different ways — "du Pont," "duPont," "DuPont," "E.I. du Pont de Nemours." They are all the same 1973 ruling.

In re E.I. du Pont de Nemours (CCPA 1973) is the one to know cold. It set out the thirteen factors examiners weigh to decide whether two marks are confusingly similar — how alike the marks look and sound, how related the goods are, how strong the earlier mark is, and more. Almost every Section 2(d) refusal runs through du Pont. Concretely: if you file DELMONT for coffee and DELMONTE already exists for canned fruit, du Pont is the grid the examiner fills in to explain the refusal.

In re Hearst (Fed. Cir. 1992) governs how examiners weigh the parts of a mark — you compare the marks as wholes, not just the part they share. Concretely: VARGA GIRL and VARGAS aren't automatically confusable just because they share "Varga"; the court faulted the refusal for giving too little weight to "girl."

In re National Data (Fed. Cir. 1985) anchors the rule that you analyze a mark as a whole even when part of it is disclaimed — you can't ignore descriptive matter just because the applicant gave up exclusive rights to it. Concretely: if your mark is FIRST CLEARANCE and "clearance" is disclaimed, the examiner still looks at the entire mark when judging confusion, not just "first."

In re N.A.D. (Fed. Cir. 1985) anchors the sophisticated-purchaser argument: when the buyers are careful professionals making considered decisions, confusion is less likely — and a consent agreement between the two owners can support withdrawing a confusion refusal. Concretely: two similar marks on hospital anesthesia equipment can coexist where the same two names on impulse-buy consumer goods could not. (You'll meet this idea again in our du Pont series — it's factor four.)

In re Coors Brewing (Fed. Cir. 2003) shapes how examiners judge whether food and restaurant services are related enough to cause confusion — the answer is "not automatically." Concretely: a beer brand and a restaurant with a similar name don't necessarily clash; the examiner needs more than "both involve food."

Beyond the top tier, a few cases surface repeatedly in the harder arguments. In re Thor Tech (TTAB 2015) deals with the weight of coexisting registrations — the fact that similar marks already live side by side on the register. Coach Servs. v. Triumph Learning (Fed. Cir. 2012) governs a mark's fame and strength. Juice Generation v. GS Enterprises (Fed. Cir. 2015) addresses crowded-field weakness — when a word is so common in an industry that no single owner gets much protection. In re i.am.symbolic (Fed. Cir. 2017) shows the other end of the scale: the marks were legally identical, so the first du Pont factor weighed heavily, and a restriction tying the goods to will.i.am did not meaningfully limit them, so it did not save the application. And In re Colonial Stores (CCPA 1968) covers the double-entendre argument, where a term that looks descriptive is saved because it also carries a second, non-descriptive meaning.

For a founder, the practical read is simple: the outcome of your refusal was substantially decided before you filed, by courts you've never heard of. For a junior practitioner, the edge is knowing that a persuasive response speaks these cases' language back to the examiner. You can read more on how these precedents get deployed in real prosecution in our insights library.

Where this series goes next

Now that you know an office action is a citation-backed letter on a deadline — and that most of it is paperwork while the substantive refusals are the real contest — the rest of the series walks the refusals themselves.

Part 2 takes on the most important one, Section 2(d) likelihood of confusion. Later parts take the other refusals one at a time: what triggers each, why examiners raise it, and what the register shows about how often it sticks. From there we open the casebook behind each refusal — the specific precedents examiners lean on — and close the series with what actually works in a response, based on how applicants have gotten refusals withdrawn.

If you want to see a specific precedent in depth, our precedent pages cover the heavy hitters — including In re du Pont — with the doctrine and the citations examiners actually use.

If you already have an office action, GleanMark's free Office Action Decoder lists the issues the examiner raised, your response deadline and, for a likelihood-of-confusion refusal, the registration the examiner cited. That breakdown helps you see which issues are paperwork and which are substantive refusals.

This analysis is based on public USPTO records and is not legal advice. Deadlines and fees checked against the USPTO's rules on October 12, 2026.