# Is Your Invention Patentable? A Screening Checklist > Screen your invention before you spend money: statutory category, Alice red flags, bar dates that may have already killed it, obviousness, enablement. Guide | Author: Lidiia Levitska | Source: Intellectual Property Law (outsideipcounsel.com) Canonical: https://outsideipcounsel.com/guides/what-is-patentable/

Quick answer: Before you pay for a professional search or an attorney's opinion, you can screen your own invention in an evening. Run six gates in this order: (1) does it fit a statutory category; (2) does it trip an Alice red flag, and if so, can you name the technical improvement; (3) have you already started a bar clock by disclosing, selling, or offering it; (4) is it honestly just a predictable combination; (5) can you teach the full scope you want to claim; (6) would the patent be detectable and enforceable. Gates 1 through 5 decide whether you can get a patent. Gate 6 decides whether you should want one. Stop at the first hard failure and save the money.

This guide is not about what the law says. It is about what you do with your own invention on a Tuesday night before you spend anything. For the underlying doctrine, what the categories are, how the courts built the judicial exceptions, and how the two-step framework actually runs, see the concept page on [patent eligible subject matter and Alice](/topics/patents/eligibility/). Read that first if the terms below are unfamiliar. What follows assumes you broadly know the rules and now want to apply them to your thing. One framing note before you start. You are not trying to prove your invention is patentable. You are trying to find the reason it is not, as cheaply and as early as possible. An examiner will find it eventually. Finding it yourself costs nothing. ## Gate 1: Does it fit a statutory category, and which one? This gate is fast and almost everyone passes it, but do it anyway because the answer shapes everything downstream. Write one sentence: "My invention is a ___." The blank has to be a **process**, a **machine**, a **manufacture**, or a **composition of matter**, per [35 U.S.C. § 101](https://www.law.cornell.edu/uscode/text/35/101). If your honest answer is "a business model," "a service," "a website," "an idea for an app," or "a discovery," you have not failed yet, but you have learned something important: you have described the *goal*, not the invention. Rewrite the sentence until the blank is a thing you could build or a sequence of steps you could perform. The useful output of this gate is not pass or fail. It is a forced act of specificity. If you cannot fill the blank without hand-waving, gates 2 and 5 are going to hurt. **Trap:** if your invention is really about how a product *looks* rather than how it works, you are in the wrong system. Ornamental appearance goes to a design patent, and the analysis is completely different. See [design vs. utility patent](/guides/design-vs-utility-patent/) before you go further. ## Gate 2: Run the Alice red-flag self-test This is the gate that kills the most software, fintech, diagnostic, and data-analytics inventions, and it is the one inventors screen for last, if at all. Do it second. **The test.** Describe your invention in one sentence with the computer removed. Not "de-emphasized." Removed. No "processor," no "server," no "database," no "using machine learning." Now look at what is left. Flag it if the remaining sentence is any of these: - **A method of organizing human activity.** Matching buyers to sellers, hedging risk, escrow, loyalty points, scheduling people, enforcing a rule about who may do what. - **A mental process.** Something a sufficiently patient person could do in their head or with pen and paper: comparing, classifying, judging, deciding. - **A mathematical relationship or formula.** An algorithm, a scoring function, a statistical model, stated as such. - **A natural correlation.** "When biomarker X is at level Y, condition Z is likely." Discovering the correlation is not the same as inventing something. - **A bare idea implemented on a generic computer.** The tell: the only thing the computer adds is speed, scale, or automation of what people already did manually. If nothing flags, this gate is likely clear and you can move on. If something flags, you have not lost. You have found the exact fight your application will have, roughly a year before an examiner tells you about it. Here is what to do about it. ### If you flagged: find the technical improvement or stop The question that decides your case is: **what is technically better, and better about what?** The answer has to be about the *technology*, not the business outcome. "Users convert 30% more" is not a technical improvement. "The index structure reduces memory writes because it is self-referential" is. Three things to do, in order: 1. **Name the improvement in one sentence, to a machine.** The framing you want is the one *Enfish* established and the [eligibility concept page](/topics/patents/eligibility/) walks through: an improvement in the functioning of the computer itself, not an abstract idea using a computer as a tool. For screening purposes the test is blunt. If you cannot articulate the improvement in one sentence without mentioning revenue, users, or convenience, you probably do not have one. 2. **Check whether the improvement is actually in your write-up.** This is where inventors lose winnable cases. The specification has to describe the technical advance with enough specificity that it reads as an engineering disclosure, not a pitch deck. If your draft says "the system optimizes routing" and never says *how*, there is nothing for anyone to point to later. Go add the how, with specifics: the data structure, the ordering, the constraint, the tradeoff you resolved. 3. **Claim structure, not aspiration.** Recite the specific components, the specific steps, the specific arrangement. Functional language at the "means for achieving good outcome" altitude is what draws the abstract-idea rejection in the first place. **One more lever worth knowing.** Because "well-understood, routine, and conventional" is a question of fact rather than something an examiner can simply assert (see [*Berkheimer* and step two](/topics/patents/eligibility/)), the drafting move is to put the evidence in your own specification: say plainly, while you are writing, what about the technique was *not* conventional and why. That is worth doing now rather than scrambling for it later. **When to just stop.** If the honest answer to "what is technically better" is "nothing, we just did an old thing on a phone," this gate is a hard fail. That is a good outcome for you. You just saved five figures. Consider whether the real asset is the execution, the brand, or the data, and read [patent vs. trade secret](/guides/patent-vs-trade-secret/). ## Gate 3: Check whether you already killed it This is the most painful gate because the failure is usually already in the past and cannot be undone. Do it before you get emotionally invested. Get out a calendar and write down every date on which any of the following happened, anywhere in the world: - You **published** it: a paper, a blog post, a conference talk, a demo video, a GitHub repo, a Kickstarter page, a trade-show booth. - You **offered it for sale** or sold it. Note carefully: a commercial offer counts even if the buyer was under an NDA. *Helsinn Healthcare S.A. v. Teva Pharmaceuticals USA, Inc.* (2019) held that a sale to a third party obligated to keep the invention confidential can still trigger the on-sale bar under the AIA. - You **publicly used** it in a way that was not secret and not experimental. - Someone **else** disclosed the same thing. Now apply the rule. [35 U.S.C. § 102(a)](https://www.law.cornell.edu/uscode/text/35/102) makes something prior art if it was patented, described in a printed publication, in public use, on sale, or otherwise available to the public before your effective filing date. Section 102(b)(1) then carves out a narrow personal grace period: a disclosure made **1 year or less** before the effective filing date is not prior art if the disclosure came from the inventor, or from someone who got it from the inventor. Read that carefully, because it is narrower than inventors think: - **It is yours, not everyone's.** It does not protect you from an independent third party's disclosure. If a stranger published the same thing yesterday, the grace period does nothing for you. - **It is a deadline, not a shield.** The moment you disclosed, a 12-month clock started. Twelve months and a day later, your own talk is prior art against you. - **It is largely a U.S. thing.** Most foreign systems apply absolute novelty: a public disclosure before filing destroys the right, with only narrow exceptions such as abuse of the applicant's rights or certain officially recognized exhibitions. If you have any international ambition, treat your first public disclosure as the end of your foreign rights unless a filing predates it. See [international patent protection](/guides/pct-international-patents/). **Decision rule.** If any of those dates is more than 12 months old and matches what you want to claim, this may be a hard fail for that subject matter, and it is a question to take to counsel immediately rather than discover later. If any of those dates is *inside* 12 months, you no longer have a screening project. You have a deadline. File something. **If nothing has been disclosed:** do not disclose anything until you have a filing date or a signed NDA. That is the single highest-value action in this entire guide. See [NDA before pitching an invention](/guides/nda-before-pitching-invention/). ## Gate 4: The obviousness honesty check Novelty is a search question and you should run one: see [how to do a patent search before you file](/guides/patent-search-before-filing/) for the actual mechanics of the free tools. This gate is different. It is the one you have to be honest with yourself about, because no search result tells you the answer. Ask this and answer it as if you were a competitor's expert witness being paid to say no: > If a competent engineer in my field had been handed the problem I set out to solve, with access to everything already public, would they have gotten here? Under [35 U.S.C. § 103](https://www.law.cornell.edu/uscode/text/35/103), the question is whether the differences between your invention and the prior art would have been obvious to a person of ordinary skill. *KSR International Co. v. Teleflex Inc.* (2007) made this harder for applicants: the Court rejected a rigid requirement of an explicit teaching or motivation to combine and endorsed a flexible, common-sense approach, noting that combining familiar elements according to known methods is likely obvious when it yields predictable results. So the words to be afraid of are **predictable** and **known**. Test your own invention against them: - Did you combine two known things? Was the result what you would have expected? That is the KSR fact pattern. - Did you substitute one known material or component for another known equivalent? - Did you scale, miniaturize, automate, or digitize something known? - Was there an obvious problem, a finite number of solutions, and you tried them? That is "obvious to try." **How to pass this gate honestly.** The strongest thing you can have is a reason the result was *not* predictable. Write down, now, while you remember: what did you expect to happen that did not? What did people in the field say would not work? What tradeoff did everyone assume was unavoidable that you broke? What failed first? Those notes are the raw material for unexpected results, teaching away, and long-felt need, and they are also the things you will not be able to reconstruct in three years. Keep them dated. **Decision rule.** If your invention is genuinely just "A plus B, and it worked exactly as you would expect," this gate is a soft fail. Not automatic, because these arguments are fact-intensive, but you should go into counsel knowing that this is where your money will go. ## Gate 5: Can you teach what you want to own? Inventors want broad claims. Broad claims require broad teaching. This gate is where those two facts collide, and it is the gate most people skip entirely. Do this exercise. Write down the scope you *want*: not the one thing you built, but the class you want to own. "Any antibody that binds here and blocks this." "Any system that does X by means of Y." Now ask: **could a skilled person in my field, reading only what I have written, make and use every member of that class without undue experimentation?** *Amgen Inc. v. Sanofi*, 598 U.S. 594 (2023), is the case to have in mind. The Supreme Court unanimously held that Amgen's claims, which covered an entire genus of antibodies defined by their function, were not enabled: if a patent claims an entire class of processes, machines, manufactures, or compositions of matter, the specification must enable a skilled person to make and use the entire class. The Court's formulation is the one to remember: **the more one claims, the more one must enable.** A roadmap that amounts to "here are some examples, now go run the same trial-and-error we did" was not enough. The screen version of that: - **Count your examples.** If you want to claim a class and you have one working embodiment, ask what makes you confident the rest of the class works. - **Look for functional claiming.** If your desired scope is defined by *what the thing does* rather than *what it is*, that is the Amgen shape. It is not automatically fatal, but it means your teaching has to carry more weight. - **Identify the trial and error.** Anywhere a reader would have to experiment to get from your disclosure to a working version, that is the gap. Ask whether that experimentation is routine or a research project. **Decision rule.** This gate rarely produces a stop. What it produces is a *scoping* decision: the honest breadth of your claim is roughly the breadth of your teaching. If those two do not match, either go do more work and generate more disclosure, or accept a narrower patent. Deciding this before drafting is much cheaper than discovering it in prosecution. For the doctrine, see the [enablement analyses](/topics/patents/enablement/). ## Gate 6: Would the patent actually be worth having? Gates 1 through 5 ask whether you *can*. This one asks whether you *should*, and it is the gate that should kill the most applications and almost never does. Three questions. **Can you detect infringement?** A patent is a right to sue. You cannot sue what you cannot see. Ask: if a competitor did exactly this, would I know? If the invention lives inside a factory, a server, or an internal process that never ships to a customer, the honest answer is often no. A claim to an invisible manufacturing step is a public teaching document handed to your competitors in exchange for a right you can never assert. That is the classic case for keeping it secret instead: see [patent vs. trade secret](/guides/patent-vs-trade-secret/). **Could you afford to enforce it?** Patent litigation is expensive and slow enough that the realistic question for most individual inventors and small companies is not "would I win" but "who would pay for the fight." If the answer is nobody, the patent's value is defensive, reputational, or transactional, in licensing, in a raise, in an acquisition, rather than in enforcement. That is a legitimate reason to file. It is just a different reason, and it should change what you spend. See [patent litigation cost and timeline](/guides/patent-litigation-cost-and-timeline/). **Does the 20-year clock match your technology?** A utility patent term runs 20 years from the earliest non-provisional filing date, and issuance typically takes years. If your field turns over every 18 months, you may be buying protection for something nobody will care about by the time you can assert it. **Decision rule.** If the invention is undetectable in a competitor's product, and you have no licensing or fundraising use for the paper, seriously consider not filing. Secrecy has no expiry, no fee schedule, and no publication. ## What to do once you have screened The screen produces one of three outcomes. Act on it. **Hard fail at gate 2, 3, or 6.** Stop. Write down why, so future-you does not re-litigate it. Redirect the budget to whatever the actual asset is: trade secret hygiene, brand, or speed. **Clean pass, nothing disclosed.** You are in the best position available. Do not disclose. Run a knockout prior-art search yourself before you pay anyone. Then go to counsel. **Clock running.** You disclosed, offered, or sold inside the last 12 months. Priority beats polish. A provisional application is the standard cheap anchor: the USPTO fee is **$325** for a large entity, **$130** for a small entity, and **$65** for a micro entity, and it buys 12 months to file a nonprovisional claiming its benefit under [35 U.S.C. § 119(e)](https://www.law.cornell.edu/uscode/text/35/119). The trap with provisionals is worth stating plainly: **a provisional only anchors what it actually describes.** It is not a placeholder for a concept. If it does not teach how to make and use the thing, it will not support the claims you eventually want, and you will have paid $65 for a date that does not attach to anything. Whatever you file, file the real disclosure. See [provisional vs. nonprovisional patent](/guides/provisional-vs-nonprovisional-patent/) and [poor man's patent](/guides/poor-mans-patent/) for the myth version of the same instinct. ### What to bring to counsel If you have run the gates, you can walk into a first meeting with something most inventors never bring: a screened invention. Bring: 1. The one-sentence category statement from gate 1. 2. Your gate 2 answer: the technical improvement, in one sentence, or an honest note that you could not find one. 3. **The date list from gate 3.** Every disclosure, offer, sale, and public use, with dates. This is the single most valuable thing you can hand an attorney, and the one they most often have to extract painfully. 4. The prior art you found yourself, and specifically the closest reference, not the most flattering one. 5. Your gate 4 notes: what was unexpected, what failed, what the field assumed. 6. The scope you want versus the scope you can teach, from gate 5. 7. Your gate 6 answer: how you would detect infringement, and what you would do about it. That list turns a $500 conversation about your idea into a $500 conversation about your strategy. See [when to hire IP counsel](/guides/when-to-hire-ip-counsel/). ## The bottom line Screening is not a shortcut around counsel. It is how you stop paying counsel to discover things you already knew. Run the gates in order, because they are ordered by how expensive the failure is to find later: category, then Alice, then your own bar dates, then obviousness, then enablement, then economics. The two that matter most are the two inventors skip. Gate 3 is the only one where the answer may already be locked in the past, so check it before you fall in love with the project. Gate 6 is the only one that asks whether the right you are buying is a right you could ever use. If you fail a gate, that is the screen working. The whole point is to fail cheaply. ## Frequently asked questions **How do I know if my invention is worth patenting?** Run six gates in order, and stop at the first hard failure. Does it fit a statutory category (process, machine, manufacture, composition of matter)? Is the heart of it an abstract idea, mental process, or natural law? Has your own disclosure, sale, or public use already started a bar clock? Is it just a predictable combination of known parts? Can you actually teach the full scope you want to claim? And would a patent be detectable and enforceable against a competitor? The first five gates decide whether you can get a patent. The last one decides whether it is worth having. **I already showed my invention publicly. Is it too late to file?** In the United States, maybe not. Under [35 U.S.C. § 102(b)(1)](https://www.law.cornell.edu/uscode/text/35/102), a disclosure made by the inventor, or by someone who got it from the inventor, one year or less before the effective filing date is not prior art against that inventor. That is a personal 12-month grace period, not a general one. It does not protect you from a third party's independent disclosure, and most foreign systems have no comparable grace period at all, so a public disclosure before filing can permanently forfeit rights in Europe and elsewhere. If you have disclosed, treat the date as a deadline and get the filing date fixed. **My invention is software. Should I even bother screening for Alice?** Yes, and it should be the first thing you look at after the category check. Write your invention down in one sentence with no mention of a computer. If that sentence describes a business practice, a mental judgment, a math relationship, or a way of organizing people, the claim is at real risk under the Alice framework and you need to identify a concrete technical improvement to the computer or the technology itself, describe that improvement in the specification with specifics, and claim it. If the only answer to "what is technically better here" is "it is faster because computers are fast," that is the answer an examiner will give you too. **Should I file a provisional application while I screen?** A provisional is a cheap way to anchor a priority date while you keep working, but it only anchors what it actually describes. The USPTO provisional filing fee is $325 for a large entity, $130 for a small entity, and $65 for a micro entity, and it gives you 12 months to file a nonprovisional claiming its benefit under 35 U.S.C. § 119(e). A thin provisional that gestures at a concept without teaching how to make and use it will not support the later claims you care about, so the cost of doing it badly is a false sense of security, not just the fee. *This guide is general legal information for educational purposes, not legal advice, and does not create an attorney-client relationship. Patent eligibility, obviousness, and enablement are intensely fact-specific, fee schedules change, and the law continues to evolve. For guidance on your particular invention, consult a patent attorney or agent licensed to practice before the patent office in your jurisdiction.* ## Frequently asked questions ### How do I know if my invention is worth patenting? Run six gates in order, and stop at the first hard failure. Does it fit a statutory category (process, machine, manufacture, composition of matter)? Is the heart of it an abstract idea, mental process, or natural law? Has your own disclosure, sale, or public use already started a bar clock? Is it just a predictable combination of known parts? Can you actually teach the full scope you want to claim? And would a patent be detectable and enforceable against a competitor? The first five gates decide whether you can get a patent. The last one decides whether it is worth having. ### I already showed my invention publicly. Is it too late to file? In the United States, maybe not. Under 35 U.S.C. § 102(b)(1), a disclosure made by the inventor, or by someone who got it from the inventor, one year or less before the effective filing date is not prior art against that inventor. That is a personal 12-month grace period, not a general one. It does not protect you from a third party's independent disclosure, and most foreign systems have no comparable grace period at all, so a public disclosure before filing can permanently forfeit rights in Europe and elsewhere. If you have disclosed, treat the date as a deadline and get the filing date fixed. ### My invention is software. Should I even bother screening for Alice? Yes, and it should be the first thing you look at after the category check. Write your invention down in one sentence with no mention of a computer. If that sentence describes a business practice, a mental judgment, a math relationship, or a way of organizing people, the claim is at real risk under the Alice framework and you need to identify a concrete technical improvement to the computer or the technology itself, describe that improvement in the specification with specifics, and claim it. If the only answer to "what is technically better here" is "it is faster because computers are fast," that is the answer an examiner will give you too. ### Should I file a provisional application while I screen? A provisional is a cheap way to anchor a priority date while you keep working, but it only anchors what it actually describes. The USPTO provisional filing fee is $325 for a large entity, $130 for a small entity, and $65 for a micro entity, and it gives you 12 months to file a nonprovisional claiming its benefit under 35 U.S.C. § 119(e). A thin provisional that gestures at a concept without teaching how to make and use it will not support the later claims you care about, so the cost of doing it badly is a false sense of security, not just the fee.