Checklist · Intellectual Property

Section 101 Eligibility After Alice: A Practical Checklist

Type
Checklist
Jurisdiction
National
Date
2026-04-30
Topic
Intellectual Property

More than a decade after Alice Corp. v. CLS Bank, software, fintech, and life-sciences claims still die on the pleadings under 35 U.S.C. § 101—yet a narrow, well-marked path to eligible claims remains open to drafters who understand the rules. This checklist walks a patent drafter or prosecutor through the Alice/Mayo two-step as the USPTO administers it (Step 2A Prong One, Step 2A Prong Two, and Step 2B), through the Federal Circuit guideposts that mark eligibility (Enfish, DDR Holdings, McRO, BASCOM), and through the recurring patterns of ineligibility to avoid (the collect-analyze-display trap, the "do it on a computer" move, the specification-claim disconnect). Each principle is translated into concrete claim-drafting, specification, and prosecution moves, including the Berkheimer factual showing, the 2019 Revised Guidance practical-application prong, and the 2024 AI examples. It maps the procedural stakes—why eligibility can be decided at the motion-to-dismiss stage—and the life-sciences overlay of Mayo and Myriad. The goal is to draft and prosecute claims that live on the right side of the line while § 101 reform remains stalled.

Section 101 Eligibility After Alice: A Practical Checklist

Intellectual Property > Patent | Published 30 April 2026 - Updated 2 June 2026 | UpLaw - uplaw.us

A software or diagnostic patent can be extinguished on a motion to dismiss—before an answer, before claim construction, before a single deposition—when a judge sets your claims beside Alice and pronounces them an "abstract idea." Eligibility turns far less on what was invented than on how the claim was written. This checklist supplies the moves that keep a claim on the eligible side of the line.

Keywords: Section 101, Alice, patent eligibility, 35 U.S.C. § 101, abstract idea, judicial exception, Mayo two-step, inventive concept, practical application, Enfish, DDR Holdings, McRO, BASCOM, Berkheimer, USPTO 2019 PEG, Electric Power Group, preemption, PERA

Patent lawyers like to say that eligibility is a threshold question. That undersells it. Section 101 is the trapdoor. A claim that would sail through novelty, nonobviousness, and written description can vanish at the pleadings stage because a district judge—reading the claim on its face, indulging every factual inference in the patentee's favor—still concludes it is "directed to" an abstract idea and adds nothing more. No prior-art search. No expert. No jury. Just a comparison to a growing library of dead patents.

The doctrine is famously slippery, but it is not lawless. Over a decade of Federal Circuit decisions has produced a recognizable set of survivors and a much larger pile of casualties, and the difference between them is remarkably consistent: survivors recite a specific technical mechanism that improves how a machine or a technological process works; casualties recite a result and gesture at a generic computer to achieve it. This checklist reduces that pattern to a sequence of decisions a drafter and prosecutor can actually run—the same two-step the USPTO applies, the guideposts that mark the eligible path, the traps that mark the cliff edge, and the factual showings that decide close cases.

For the strategic companion to this operational checklist—portfolio design, prosecution tactics across a family, and the business case for filing at all—pair it with the Section 101 strategy guide for software and business-method inventions. For the broader procedural context of answering an examiner, use the checklist for responding to a patent office action.

How to use this checklist

Work it in order. Phase 0 orients you to the doctrine so the later checkboxes make sense. Phases 1 through 3 track the USPTO's examination framework step by step—that is the analysis an examiner will run and the analysis a district court will echo. Phases 4 and 5 are the drafting payoff: how to write claims and specifications that pass. Phase 6 turns the USPTO's own guidance into arguments. Phase 7 handles portfolio strategy and the reform picture. Two short interludes address the life-sciences overlay and the procedural stakes, because eligibility law is not one doctrine but two—computer-implemented and life-sciences inventions run on parallel but distinct tracks.

A note on scope. This is a drafting-and-prosecution resource, written from the applicant's side of the desk. If you are attacking a patent rather than defending one, read every checkbox as a target: the same features that make a claim survive are the features whose absence you will exploit.

Phase 0 — Orient: the statute, the exceptions, and the engine underneath

Before the first checkbox, fix the architecture in your mind.

Section 101 is the first of several gates. It asks a categorical question—is this the kind of thing the patent system covers at all?—before the Patent Act's other requirements ask whether this particular invention is new (§ 102), nonobvious (§ 103), and adequately described and enabled (§ 112). The statute itself is generous. It extends patent protection to "any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof." 35 U.S.C. § 101; see 35 U.S.C. § 100 (defining "process" to include a new use of a known process, machine, manufacture, composition of matter, or material). The Supreme Court has read that language expansively, quoting the 1952 Act's legislative history for the proposition that Congress meant patentable subject matter to include "anything under the sun that is made by man." Diamond v. Chakrabarty, 447 U.S. 303, 309 (1980).

The judicial exceptions are the whole fight. Against that generous text, the Court has long recognized three implied carve-outs: laws of nature, natural phenomena, and abstract ideas. Diamond v. Diehr, 450 U.S. 175, 185 (1981). These are judge-made. You will not find them in the statute; you will find them in a century of decisions worried about one thing above all—preemption. The animating concern is that a patent on a fundamental building block of science or commerce would "tie up" that building block and choke off the innovation the patent system exists to promote. Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 72 (2012). Keep preemption in view throughout; it is not a separate test, but it is the policy engine, and a claim that plainly leaves the field open to others is a claim that answers the deepest objection to its own eligibility.

Two Supreme Court decisions supply the modern test. Mayo (a natural-law case in the life sciences) and Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014) (an abstract-idea case in software), together frame a two-step inquiry that governs every technology:

  1. Step one: Are the claims directed to a judicial exception—a law of nature, a natural phenomenon, or an abstract idea?
  2. Step two: If so, do the claims contain additional elements—considered individually and as an ordered combination—that amount to "significantly more" than the exception itself? The Court calls this the search for an inventive concept: claim elements "sufficient to ensure that the patent in practice amounts to significantly more than a patent upon the [ineligible concept] itself." Alice, 573 U.S. at 217–18 (quoting Mayo, 566 U.S. at 72–73).

The Court deliberately refused to define "abstract idea." In Alice, it declined to "labor to delimit the precise contours of the 'abstract ideas' category," instead reasoning by analogy: the risk-hedging claims of Bilski v. Kappos, 561 U.S. 593 (2010), and the settlement-risk claims of Alice were both "fundamental economic practice[s] long prevalent in our system of commerce," and therefore abstract. Alice, 573 U.S. at 219–20. That analogical method is not a bug; it is the operating system. As the Federal Circuit has candidly acknowledged, there is "no single, succinct, usable" definition of an abstract idea, so courts compare a challenged claim to the claims held eligible and ineligible in earlier cases. Amdocs (Israel) Ltd. v. Openet Telecom, Inc., 841 F.3d 1288, 1294 (Fed. Cir. 2016). Which is exactly why a working knowledge of the leading cases—not a formula—is what separates drafters who win from drafters who guess.

A few coordinates from the older cases repay memorizing, because the modern framework is built on them:

  • Gottschalk v. Benson, 409 U.S. 63 (1972), held a claim to a mathematical algorithm for converting binary-coded decimals to pure binary ineligible, because it would in practice preempt the algorithm itself.
  • Parker v. Flook, 437 U.S. 584 (1978), held ineligible a claim reciting a formula for updating "alarm limits" in catalytic conversion; the point of novelty was the math, and the surrounding steps were "well known." There was no inventive concept in the application.
  • Diamond v. Diehr, 450 U.S. 175 (1981), came out the other way: a process for curing rubber that used the Arrhenius equation to time the mold's opening was eligible, because the claim did not seek to patent the equation but to improve a concrete, physical industrial process. The steps were not "obvious, already in use, or purely conventional."

The through-line from Flook to Diehr to Alice is the same lesson the checklist will hammer: reciting an ineligible concept and then bolting on conventional, generic activity does not make it eligible; genuinely applying that concept to improve a specific technology does.

For the wider frame—how § 101 sits alongside §§ 102, 103, and 112 in the life of a patent—see how U.S. patents work: a comprehensive practical guide. For the plain-language version founders ask about, see patents for inventors and founders: frequently asked questions.

Phase 1 — Run the two-step the way the USPTO runs it

The USPTO administers Alice/Mayo through a structured flowchart that adds one crucial refinement the courts later borrowed. Run every claim through it explicitly; do not skip steps, because the answer often turns on which step you are actually in.

  • [ ] Step 1 — statutory category. Confirm the claim recites a process, machine, manufacture, or composition of matter. This is usually easy, but not always: a claim to a bare signal—a "transitory, propagating signal" carrying data—falls outside all four categories and is ineligible for that reason alone, independent of Alice. In re Nuijten, 500 F.3d 1346 (Fed. Cir. 2007); see Mentor Graphics Corp. v. EVE-USA, Inc., 851 F.3d 1275, 1294 (Fed. Cir. 2017). Draft "computer-readable medium" claims as non-transitory to close this trap.

  • [ ] Step 2A, Prong One — does the claim recite a judicial exception? For abstract ideas, the USPTO recognizes only three enumerated groupings (MPEP § 2106.04(a)):

    • mathematical concepts (mathematical relationships, formulas, calculations);
    • certain methods of organizing human activity (fundamental economic practices, commercial and legal interactions, managing personal behavior or relationships); and
    • mental processes (evaluations, judgments, and observations that can be performed in the human mind or with pen and paper). If the claim does not recite subject matter within one of these buckets—or a law of nature or natural phenomenon—the examiner should find it eligible without further analysis, absent a rare, Technology-Center-Director-approved exception. This enumerated-groupings approach is a real gift to applicants: it disciplines examiners who would otherwise pitch the "abstract idea" at whatever altitude defeats the claim.

  • [ ] Step 2A, Prong Two — is the exception integrated into a practical application? This is the USPTO's key addition, introduced in the 2019 Revised Patent Subject Matter Eligibility Guidance (2019 PEG), 84 Fed. Reg. 50 (Jan. 7, 2019), and now embedded in MPEP § 2106.04(d). Ask whether additional claim elements apply the exception in a way that imposes a meaningful limit—for example, by improving the functioning of a computer or another technology, or by using the exception in conjunction with a particular machine or transformation. If the claim integrates the exception into a practical application, it is eligible and the inquiry ends. Two features of Prong Two make it the drafter's best friend: it can rest on elements that are routine and conventional (unlike Step 2B), and it lets you win before ever reaching the harder inventive-concept question.

  • [ ] Step 2B — does the claim add "significantly more"? Only if the claim survives to here. Ask whether the additional elements, individually and as an ordered combination, supply an inventive concept—something beyond "well-understood, routine, [and] conventional" activity. Mayo, 566 U.S. at 73; Alice, 573 U.S. at 225. A claim that does no more than require a generic computer to perform generic computer functions fails. Alice, 573 U.S. at 225.

  • [ ] Hold preemption in view the whole way through. It is not a checkbox of its own, but a narrow, specific claim that visibly leaves competitors free to solve the same problem differently is your most persuasive theme at every step. The converse is a warning: the Federal Circuit has held that the absence of complete preemption does not by itself confer eligibility. Ariosa Diagnostics, Inc. v. Sequenom, Inc., 788 F.3d 1371, 1379 (Fed. Cir. 2015). Preemption helps you; its absence does not save you.

Where the two frameworks converge and diverge. The USPTO's Prong Two ("practical application") maps onto the courts' step-one inquiry ("directed to"): both ask whether the claim, read as a whole, is really about the exception or about a specific application of it. The USPTO's Step 2B maps onto the courts' step two. The vocabulary differs; the substance is close enough that a claim engineered to pass the USPTO flowchart is a claim built to survive an Alice motion. Note one authority limit: the Federal Circuit is "not bound" by USPTO guidance and has said so expressly. In re Rudy, 956 F.3d 1379, 1382–83 (Fed. Cir. 2020); Cleveland Clinic Found. v. True Health Diagnostics LLC, 760 F. App'x 1013, 1020 (Fed. Cir. 2019). Use the guidance as a sword before the examiner; do not treat it as controlling before a court.

Phase 2 — Win at Step 2A: frame the invention as a technological improvement

Most eligible software claims win at step one / Prong Two, not step two. The reason is Enfish: a claim can be eligible because it is not directed to an abstract idea in the first place—because it improves the computer itself.

  • [ ] Name the technical problem and the mechanism that solves it. Before drafting, write one sentence: "Prior systems could not do X because of technical limitation Y; this invention does X by means of specific mechanism Z." If you cannot fill in a concrete Z, you do not yet have an eligible claim—you have a result in search of a mechanism.

  • [ ] Frame the invention as an improvement in how the computer operates, not a generic computer used to implement an idea. In Enfish, LLC v. Microsoft Corp., 822 F.3d 1327 (Fed. Cir. 2016), the claims recited a "self-referential" database table—a table that indexes its own columns and rows in a single table rather than the conventional multi-table relational model. The court held the claims eligible at step one because they were "directed to an improvement of an existing technology," achieving faster searching and more efficient memory use. The claimed advance was "a specific type of data structure designed to improve the way a computer stores and retrieves data." That is the template: an improvement to the machine, recited in the claim.

  • [ ] Anchor the claim to a problem "necessarily rooted in computer technology," solved in an unconventional way. DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245 (Fed. Cir. 2014), remains the paradigm. The claims addressed a problem unique to the internet—retaining website visitors who click a third-party advertisement, who would ordinarily be whisked away to the advertiser's site—by generating a hybrid page that combined the host site's "look and feel" with the third party's product data. The solution was "necessarily rooted in computer technology in order to overcome a problem specifically arising in the realm of computer networks," and it did not merely recite a pre-internet business practice on a computer. Eligible.

  • [ ] Recite particular rules or a defined process that replaces subjective human judgment—and does not claim every way of doing the task. McRO, Inc. v. Bandai Namco Games America Inc., 837 F.3d 1299 (Fed. Cir. 2016), upheld claims to automated lip-synchronization of animated characters. The claims recited specific rules—defined relationships among phonemes, timing, and morph weights—that produced results a skilled animator previously achieved by hand and intuition. Crucially, the claims did not preempt all rules-based lip-sync; they claimed a particular genus of rules. The court warned against overgeneralizing the claim to a high level of abstraction "untethered from the language of the claims." That warning is a drafting instruction: the specificity of your rules is the eligibility.

  • [ ] Police the level of abstraction—pull the claim down from the label to its limitations. Every invention, the Court reminds us, embodies a natural law or abstract idea "at some level," so a step-one inquiry that stops at a high-level label proves too much. Alice, 573 U.S. at 217; Mayo, 566 U.S. at 71. When an examiner or an accused infringer characterizes your claim as "the abstract idea of organizing information," insist on the actual limitations: the specific data structure, the specific rules, the specific ordered steps. The gravitational pull of the doctrine is always upward, toward a broader and vaguer characterization; your job is to keep the analysis anchored to the words of the claim.

Worked example — the eligible framing. Suppose Verdant Grid, Inc. (a hypothetical) invents a way to keep an electrical microgrid stable when rooftop-solar output swings unpredictably. The temptation is to claim "a method of balancing supply and demand on an electrical grid using a computer"—a fundamental economic-and-engineering practice, generically computerized, and dead on arrival. The eligible version claims the mechanism: a controller that samples inverter phase-angle data at a defined sub-cycle interval, applies a specified damping algorithm to predict oscillation onset, and reconfigures the physical switching topology of the distribution feeders before the oscillation propagates—reciting the particular sampling rate, the particular control rule, and the physical reconfiguration. That claim improves how a real technological system operates (Enfish/DDR), recites particular rules rather than every way to stabilize a grid (McRO), and ends in a concrete physical action rather than a displayed number. It is written to win at Prong Two.

For the deeper doctrinal survey behind these guideposts—and how they play out across a portfolio—see protecting software and business-method inventions after Alice, and for the family-level view of recent developments, pivotal patent decisions of late 2024 and their IP-strategy fallout.

Phase 3 — Win at Step 2B: an inventive concept grounded in fact

If the claim survives to Step 2B, the game changes from framing to evidence. The pivotal move here is procedural: whether a claim element is "well-understood, routine, and conventional" is a question of fact, and that fact must be proven.

  • [ ] Point to a non-conventional, non-generic arrangement of known pieces. An inventive concept need not be a brand-new component. In BASCOM Global Internet Services, Inc. v. AT&T Mobility LLC, 827 F.3d 1341 (Fed. Cir. 2016), the individual elements—a local computer, a remote ISP server, filtering software—were all conventional. The inventive concept lived in the ordered combination: installing a customizable content filter at a specific location on the network (the ISP server) that allowed per-user filtering configurations while remaining remote from the user's device. "An inventive concept can be found in the non-conventional and non-generic arrangement of known, conventional pieces." Id. at 1350. When your components are individually ordinary, make the architecture the invention.

  • [ ] Recite a specific, unconventional technical step or structure—not a result. The difference between BASCOM (eligible) and the casualties of Phase 4 is that BASCOM claimed a particular configuration, while the casualties claimed the outcome and left the method a black box.

  • [ ] Demand Berkheimer support for any conventionality assertion. Berkheimer v. HP Inc., 881 F.3d 1360, 1368 (Fed. Cir. 2018), holds that whether claim elements are well-understood, routine, and conventional is a factual question that can defeat summary judgment—and, at the examination stage, that an examiner cannot simply declare an element conventional. The USPTO's Berkheimer memorandum, now in MPEP § 2106.05(d), requires the examiner to support a conventionality finding with one of four categories of evidence:

    1. an express statement or admission in the specification (or by the applicant) that the element is well-understood, routine, and conventional;
    2. a citation to a court decision noting the well-understood, routine, conventional nature of the element;
    3. a citation to a printed publication demonstrating the same; or
    4. a statement that the examiner is taking official notice—available only when the conventionality is so notorious as to be capable of instant and unquestionable demonstration.

  • [ ] Traverse bare official notice to force an affidavit. If an examiner asserts conventionality by official notice, traverse it. Under 37 C.F.R. § 1.104(d)(2) and MPEP §§ 2106.07(b), 2106.05(d), a proper traversal obliges the examiner to either withdraw the finding or support it with documentary evidence (or an affidavit). Many § 101 rejections rest on unsupported conventionality assertions; a disciplined Berkheimer traversal is one of the highest-yield moves in prosecution.

  • [ ] Keep conventionality distinct from novelty. These are different questions, and conflating them cuts both ways. On one hand, Mayo warns that an inventive concept requires more than a newly discovered natural law applied with conventional steps. On the other hand, subject matter can be disclosed in the prior art without being routine and conventional—novelty and conventionality are not synonyms, and an examiner who points to a single prior-art reference has not thereby shown that an element is "well-understood, routine, and conventional" to the field. See Berkheimer, 881 F.3d at 1369. Do not let a § 102/§ 103 argument masquerade as a § 101 conclusion.

Worked example — manufacturing an inventive concept from ordinary parts. CipherRoute, LLC (hypothetical) claims a data-transfer method using a router, a public network, and standard AES encryption—each element indisputably conventional. Claimed as "encrypting data and sending it over the internet," it is a mental/economic abstraction on generic hardware. Claimed as an ordered combination—fragmenting each file into shards, routing each shard over an independently selected path chosen in real time by a specified latency-and-entropy heuristic, and reassembling only at an endpoint holding a threshold set of shards—the claim recites a specific, non-conventional arrangement of conventional pieces, in the mold of BASCOM. If the examiner calls the sharding-and-routing arrangement "conventional," the applicant demands Berkheimer evidence; absent a specification admission, a case, or a publication, the assertion cannot stand.

Phase 4 — Draft to match the survivors, not the casualties

Phases 2 and 3 tell you what wins. This phase inventories what loses, because most § 101 fatalities are self-inflicted at the drafting table.

  • [ ] Put the technical improvement in the claim, not just the specification. This is the single most common fatal error. The Federal Circuit repeatedly refuses to import an unclaimed improvement from the specification to rescue a claim. Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1322 (Fed. Cir. 2016) (a claim's eligibility "cannot be furnished by the specification alone"; the inventive concept must appear "in the claims"). If your best eligibility argument lives only in the written description, you have written the wrong claim.

  • [ ] Recite concrete technical limitations, not result-oriented language. Results-focused, functional claims that "effectively cover any solution to an identified problem" are frequently ineligible. Interval Licensing LLC v. AOL, Inc., 896 F.3d 1335, 1345 (Fed. Cir. 2018); Affinity Labs of Texas, LLC v. DirecTV, LLC, 838 F.3d 1253, 1265 (Fed. Cir. 2016). Replace "determining," "optimizing," and "analyzing" as claim verbs-of-art with the specific data structures, algorithmic rules, interface designs, and architectures that achieve the result.

  • [ ] Avoid the collect → analyze → display pattern with no specified method. Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350 (Fed. Cir. 2016), is the canonical casualty. The claims gathered power-grid data from multiple sources, analyzed it, and displayed the results—useful, even valuable, but ineligible, because "collecting information, analyzing it, and displaying certain results of the collection and analysis" is an abstract idea when the claim specifies no particular technical means for the analysis. If your claim can be summarized as "gather data, crunch it, show the answer," it is on the wrong side of the line—unless you have recited how the crunching happens at a technical level.

  • [ ] Avoid the "do it on a computer" trap. Taking a longstanding human or commercial practice and reciting its performance "on a computer," "over the internet," or "using a processor" adds no inventive concept. Intellectual Ventures I LLC v. Capital One Bank (USA), 792 F.3d 1363, 1367–70 (Fed. Cir. 2015) (tracking and adjusting budgets, and tailoring web content to users, held abstract). The generic computer is a tool, not an invention.

  • [ ] Do not rely on the "draftsman's art." Recasting an ineligible method as a system, a device, or a computer-readable medium will not save it. In Alice itself, the system and media claims rose and fell with the method claims, because "the system claims are no different from the method claims in substance." Alice, 573 U.S. at 226. File your method, system, and medium claims by all means—but earn eligibility on the merits of what they recite, not the category label.

  • [ ] Keep claims narrow enough not to preempt an entire field. A dependent-claim strategy that layers in increasingly specific technical limitations gives you fallback positions and directly answers the preemption concern. Breadth is a liability at § 101 in a way it is not at § 102 or § 103.

Worked example — the casualty, and its rescue. AeroSense, Inc. (hypothetical) claims "a method comprising: receiving sensor readings from an aircraft engine; analyzing the readings to detect an anomaly; and displaying an alert." That is Electric Power Group wearing a flight suit—collect, analyze, display, no mechanism—and it will not survive a motion to dismiss. The rescue does two things: it specifies the analysis (a defined residual-signal model that compares vibration spectra against a rotor-order reference and flags a divergence exceeding a stated threshold within a defined frequency band) and it ends in a concrete action (automatically derating engine thrust and scheduling the affected module for inspection), not a display. Now the claim recites a specific technical method and integrates it into a practical application—an operational change to a physical machine—rather than a bare result.

Phase 5 — Build a specification that supports the claimed improvement

The specification does not save an under-drafted claim (Phase 4), but a thin specification can sink a well-drafted one—and a rich specification is your ammunition at Step 2A and your evidence at Step 2B.

  • [ ] Describe the technical problem, the prior art's specific limitations, and the measurable advantage the invention delivers—faster searching, a smaller memory footprint, improved security, reduced latency, or a result not previously achievable. Concrete, quantified advantages ("reduces index size by an order of magnitude") give a court something to hold onto at step one, exactly as the Enfish specification's efficiency statements did.

  • [ ] Provide enough technical detail to support the claimed improvement—architecture diagrams, data-structure definitions, algorithm steps, pseudocode, flowcharts. This detail underwrites both eligibility and the separate § 112 requirements of written description and enablement, which matter acutely for machine-learning inventions where a "black box" model risks a disclosure failure. (For the § 112 dimension of AI inventions and inventorship, see AI and inventorship: a global comparison of machine contributions to patents.)

  • [ ] Make the specification's improvement statements track the claim limitations. The disconnect to avoid: a specification that lavishly describes a technical advance the claims do not actually recite. That mismatch invites a court to find the claims "directed to" the abstract idea while the improvement sits, unclaimed and unavailing, in the description. Align them. Every asserted technical advantage should have a corresponding claim limitation, and every meaningful claim limitation should be explained in the specification.

  • [ ] Seed Berkheimer facts affirmatively—and avoid seeding admissions against yourself. A specification that explains why a particular arrangement is unconventional builds the factual record for Step 2B. Conversely, boilerplate that describes core components as "conventional," "well-known," or "standard" can be quoted back at you as a specification admission under MPEP § 2106.05(d). Describe the environment as conventional if you must; never describe the invention's distinguishing elements that way.

For the upstream discipline that feeds a strong specification—capturing mechanism, embodiments, and alternatives from the inventor—see drafting an invention disclosure: a practical checklist and the longer-form drafting an invention disclosure that maximizes patent scope.

Phase 6 — Use USPTO guidance as a sword

The USPTO's guidance is more applicant-friendly than the raw case law, and in prosecution it is often controlling in practice even if not in law. Deploy it.

  • [ ] Invoke the practical-application prong (2019 PEG, MPEP § 2106.04(d)). Show that the claim improves the functioning of a computer or another technology, or applies the exception with a particular machine or transformation. Remember the doctrinal advantage: at Prong Two you may rely on elements that are routine and conventional, because the question is integration, not inventiveness. A claim that reconfigures a physical device, controls a real-world process, or improves computer capability integrates the exception even if each recited component is ordinary.

  • [ ] Mine the USPTO examples for a matching fact pattern. The guidance ships with worked examples that examiners are trained on:

    • Example 39 (a method of training a neural network for facial detection using an expanded training set and an iterative false-positive-minimization program) is deemed eligible at Step 2A Prong One because it recites no mathematical concept, mental process, or method of organizing human activity at all. It is a powerful analogy for training-pipeline claims.
    • Example 46 (livestock management) draws the line sharply: a claim that gathers animal data, analyzes it for aberrant behavior, and displays the result is ineligible; the same claim that adds controlling a sorting gate to physically separate the flagged animals is eligible—even though operating a gate is itself routine—because the physical action integrates the exception into a practical application.

  • [ ] For AI and machine-learning claims, map onto the 2024 AI guidance update. The USPTO's July 2024 update, 89 Fed. Reg. 58128 (July 17, 2024), added AI-focused Examples 47–49 (covering, respectively, a neural-network hardware application for anomaly detection, AI-based speech-signal processing, and AI-assisted analysis of medical data). The recurring lesson tracks Example 46: an AI claim that ends in a concrete technological action or improvement integrates the abstraction; one that ends in a bare prediction or a displayed classification often does not. Claim the specific technical improvement, and where the model's output drives a downstream physical or technical process, recite that step.

  • [ ] Respect the AI limit the courts have drawn. Merely applying conventional machine-learning techniques on generic equipment to a new data environment is not eligible. Recentive Analytics, Inc. v. Fox Corp., 134 F.4th 1205 (Fed. Cir. 2025), held ineligible claims that used generic machine-learning models to generate broadcast schedules and network maps, reasoning that applying established ML methods to a new field of use—without any claimed improvement to the machine-learning technology itself—is a quintessential abstract idea. The takeaway for drafters: do not claim "iteratively training a model" in the abstract; claim the specific technical advance—a new architecture, a novel training method, a particular way the model interacts with hardware or a physical process.

  • [ ] For life-sciences claims, run the "markedly different characteristics" analysis. When the claim involves a product of nature, MPEP § 2106.04(b)–(c) asks whether the claimed product has characteristics—structure, function, or properties—markedly different from its naturally occurring counterpart. A claim can be based on a natural phenomenon without being directed to it, provided it recites something markedly different or a genuine application. The life-sciences interlude below explains how this plays out.

Interlude A — The life-sciences overlay: laws of nature and products of nature

The abstract-idea line dominates software, but § 101 has a parallel life in biotech and diagnostics, and the two Supreme Court anchors sit on the natural-law and natural-phenomenon side of the doctrine. The drafting lessons rhyme with the software rules, but the vocabulary and the leading cases differ.

The two Supreme Court anchors. Mayo held ineligible a diagnostic-and-dosing method that correlated the blood concentration of thiopurine metabolites with drug efficacy and toxicity: the correlation was a law of nature, and the surrounding steps—"administering" the drug and "determining" the metabolite level—were conventional activity that added nothing inventive. Mayo, 566 U.S. at 77–79. Association for Molecular Pathology v. Myriad Genetics, Inc., 569 U.S. 576 (2013), held that isolated, naturally occurring DNA is an unpatentable product of nature—Myriad "did not create or alter" the genetic information of the BRCA genes—while complementary DNA (cDNA), which is synthesized with the non-coding introns removed, is eligible because it is "not naturally occurring." Myriad, 569 U.S. at 590–95.

The Federal Circuit's eligible column—method-of-treatment claims. The reliable survivors are specific treatment methods:

The Federal Circuit's ineligible column—diagnostics. Claims that detect a natural correlation with conventional techniques keep failing:

The method-of-preparation escape hatch. Where a claim recites a new and useful process that exploits a natural discovery rather than claiming the discovery itself, it can be eligible. Rapid Litigation Management Ltd. v. CellzDirect, Inc., 827 F.3d 1042 (Fed. Cir. 2016) (a method of cryopreserving hepatocytes through multiple freeze-thaw cycles was eligible—the inventors did not claim the cells' natural ability to survive freezing, but a new process built on that discovery); Illumina, Inc. v. Ariosa Diagnostics, Inc., 967 F.3d 1319 (Fed. Cir. 2020) (a method of preparing a fetal-DNA-enriched fraction by size-selective separation was eligible as a preparation method, not a diagnostic correlation).

The drafting lesson for life sciences. It is the software lesson in a lab coat. Claim the application, not the correlation. Recite specific patients, specific compounds, specific doses, specific process steps, and where possible a physical transformation or a concrete treatment step—the biotech analog of "controlling the sorting gate." For the deeper treatment of these authorities, the USPTO's "markedly different characteristics" framework in MPEP § 2106.04(b)–(c) is the examination-side companion.

Interlude B — The procedural stakes: why eligibility is decided early, and how to fight the timing

Eligibility is not only a substantive question; it is a timing weapon, and understanding the procedure is half the battle.

Eligibility is a question of law based on underlying facts. Aatrix Software, Inc. v. Green Shades Software, Inc., 882 F.3d 1121, 1128 (Fed. Cir. 2018). That formulation is the patentee's best friend at the pleadings stage. Because the "well-understood, routine, and conventional" inquiry is factual, plausible factual allegations in a complaint that specified claim elements were unconventional can preclude dismissal under Rule 12(b)(6). Id. at 1125; Cellspin Soft, Inc. v. Fitbit, Inc., 927 F.3d 1306, 1317–18 (Fed. Cir. 2019). The practical instruction for a patent owner facing a § 101 motion: plead the facts—allege, specifically and plausibly, why the claimed arrangement was not routine or conventional as of the priority date, ideally tethered to the specification.

Claim construction can be a gating fact. If resolving the motion depends on a disputed claim term, the court must either adopt the non-movant's construction or actually construe the term before deciding eligibility. Aatrix, 882 F.3d at 1125; MyMail, Ltd. v. ooVoo, LLC, 934 F.3d 1373, 1379–80 (Fed. Cir. 2019). A patentee can sometimes defeat an early motion simply by surfacing a genuine construction dispute that bears on what the claim covers.

Summary judgment requires no genuine factual dispute. A court may resolve eligibility on summary judgment only where there is no genuine issue of material fact on conventionality. Berkheimer, 881 F.3d at 1368. This is the litigation face of the same Berkheimer factual showing you demanded in prosecution—the doctrine is symmetric, and the evidence you build during prosecution (or fail to) echoes for the life of the patent.

The strategic upshot. For applicants and patent owners, the procedural doctrine rewards two habits: (1) drafting specifications that affirmatively establish unconventionality (Phase 5), because those statements become the plausible factual allegations that survive a motion; and (2) pleading eligibility facts with particularity when litigation comes. For accused infringers, the calculus is the mirror image—move early, before those facts can be developed, and characterize the claim at the highest defensible level of abstraction. The full life-cycle mechanics of a patent case, including the § 101 motion's place in it, are mapped in litigating patent infringement: a roadmap and research guide and, from the small-business vantage, a patent-focused field guide to federal civil litigation. For the substantive infringement and defense framework once a claim clears § 101, see patent infringement, claims, and defenses.

Phase 7 — Portfolio and reform-aware strategy

Eligibility risk is not a reason to abandon the asset; it is a reason to build redundancy and to draft with an eye on where the law might go.

  • [ ] Layer protection where eligibility is uncertain. A software or algorithm-heavy innovation rarely lives on patents alone. Trade secret protects the parts you can keep confidential (a training-data pipeline, a tuning heuristic, a weighting scheme); copyright protects the expression of the code; design patents protect the ornamental UI. The four regimes are complementary, and a mature IP program assigns each layer to the tool that fits it. See how the law protects software: copyright, patents, trade secrets, and contract rights and the side-by-side comparison of copyright, trademark, patent, and trade secret. For app-specific layering, see protectable assets in a mobile app: an IP layer-by-layer guide.

  • [ ] Sequence the filings. File a provisional early to lock priority, then use the year to develop the technical detail and the eligible framing before the non-provisional. Continuation practice lets you keep a family pending so you can draft new claims against the § 101 doctrine as it evolves—a hedge worth its cost when the doctrine is this unsettled. The full prosecution arc is mapped in prosecuting a patent application: a roadmap and research guide.

  • [ ] Coordinate § 101 with the other rejections. An eligibility argument and an obviousness argument can pull in opposite directions—emphasizing that an element is unconventional (good for § 101) while insisting the combination was not obvious (good for § 103) requires care to avoid admissions that boomerang. Sequence and phrase responses so a concession on one front does not arm the examiner on another. For the companion analysis, see defeating a Section 103 obviousness rejection: a practical checklist and the fuller guide to obviousness under Section 103.

  • [ ] Prosecute under current doctrine while watching for reform. Alice remains the governing law as of mid-2026. The Patent Eligibility Restoration Act (PERA)—reintroduced in the Senate in 2025 (S. 1546) by Senators Coons and Tillis—would legislatively eliminate the judicial exceptions and replace them with a narrower set of statutory exclusions, an explicit response to the diagnostic-and-software casualties this checklist catalogs. It has not been enacted, and prior versions stalled. Frame every response under the law as it stands; do not draft claims that depend on a bill that may never pass. But keep continuations alive so that, if the ground shifts, you can move with it.

Common mistakes, distilled

A one-page eligibility scorecard

Run any draft claim through these five questions. A "no" is a red flag; two or more is a rewrite.

  1. Mechanism. Does the claim recite a specific technical means—a data structure, an algorithmic rule, an architecture, a physical step—rather than only a desired result? (Enfish, Electric Power Group.)
  2. Improvement. Does that mechanism improve how a computer or another technology operates, or drive a concrete real-world action? (Enfish, DDR, Example 46's sorting gate.)
  3. Specificity / preemption. Does the claim leave others free to solve the same problem a different way, or would it capture every solution? (McRO, Interval Licensing.)
  4. Location. Does the eligibility-conferring feature appear in the claim, not merely in the specification? (Intellectual Ventures I v. Symantec.)
  5. Factual support. If the examiner or a court calls an element conventional, do you have specification statements and plausible allegations that put that in genuine dispute? (Berkheimer, Aatrix.)

Primary authority

Related UpLaw Resources

Related Guides

Related Checklists

Related Toolkits

Related Articles

This checklist is general information, not legal advice. Section 101 doctrine is unsettled and fast-moving; consult a registered patent practitioner before acting on any point above.