The question you have to answer before you launch

Northgate Pay is six months from launching its mobile payment app, and the team is proud of it. Users will pay in stores by tapping a phone, authenticate with their face, and split a dinner bill in a few taps. The engineering is finished, the marketing is booked, the Series B closed last quarter, and the investors are watching the calendar. There is just one question no one in the building has answered out loud: does the app infringe somebody else's patent?

It is an uncomfortable question precisely because the stakes are so high. A company that launches a product reading on a valid patent can be enjoined from selling it at all under 35 U.S.C. § 283, ordered to pay damages that for a successful product climb into the hundreds of millions under § 284, made to cover the patent owner's attorneys' fees in an "exceptional" case under § 285—and, if the infringement is found willful, watch those damages tripled. The payments space happens to be one of the most heavily patented corners of the economy, crowded with portfolios held by banks, card networks, technology giants, and patent-assertion entities that make their living enforcing patents rather than building products. For a launch carrying years of work and millions of dollars, going to market blind is not bold. It is reckless. The responsible course is to look before leaping, through what patent lawyers call freedom-to-operate analysis.

Freedom-to-operate analysis—FTO, sometimes called "patent clearance"—is the process of identifying the patents a proposed product might infringe and assessing the risk each one poses. It answers a single, concrete question: can we make, use, sell, offer to sell, or import this product in this market without infringing someone else's valid, in-force patent? That question is different from, and forever confused with, patentability, which asks whether your invention is novel and non-obvious enough to earn its own patent. The distinction is not academic, and getting it wrong has sunk companies. Northgate can hold a fistful of its own patents on clever features of its app and still infringe third-party patents on other features. Owning a patent gives you a sword—the right to exclude others or collect royalties—but it gives you no shield. Freedom to operate is the shield, and it has to be earned on its own terms. For the foundational vocabulary this article assumes, our patent basics guide is a useful primer, and for the infringement doctrines in litigation depth, see what constitutes patent infringement: claims and defenses.

A word on scope before we start. FTO sits inside a slightly larger activity that practitioners call a patent landscape study—a survey of the patents and pending applications in a technology field, which serves competitive intelligence as much as clearance. The landscape study is where the searching and triage live; the FTO opinion is the deeper, claim-by-claim analysis of the handful of patents the landscape study flags as dangerous. We will move through both as a single continuous workflow, because in practice they are.

Step one: define exactly what you are clearing

Every FTO analysis begins with a deceptively simple question—what, precisely, are you clearing?—and the honest answer demands far more rigor than a marketing description supplies, because infringement turns on specific technical features, not product categories. A patent claiming "a beverage container with a resealable lid comprising a hinged flap and a locking tab" is not infringed by every resealable container, only by one with that particular hinged-flap-and-locking-tab structure. The same precision is owed to Northgate.

"A mobile payment app" is useless input for FTO. The analysis needs a technical specification: how the app authenticates users (face, PIN, passkey), how it talks to payment networks (which protocols, what tokenization and encryption), how transaction data is stored and processed (on the device or in the cloud, in which data structures), what specific user-interface gestures it relies on, and how it touches the phone's hardware—the NFC chip for tap-to-pay, the camera for QR scanning, the secure element or trusted enclave that holds credentials. Each of those features can implicate a different family of patents, so the FTO scope has to span all of them. Counsel cannot do this alone from a slide deck; the practice notes are emphatic that FTO requires patent counsel to meet and maintain open communication with the engineering team, because the specification need not be final—FTO usually runs in parallel with development—but it must be concrete enough to lay alongside a patent claim and compare element by element.

Two further dimensions frame the scope. The first is geography. Patent rights are territorial: a U.S. patent reaches only activity in the United States, so the analysis must be run separately for every country where the product will be made, used, sold, offered, or imported. For a true global launch that is an enormous undertaking, usually triaged by market size, the local enforcement climate, and business strategy. Northgate is launching first in the United States and will start there, expanding the analysis as it enters new markets. The second dimension is time. A U.S. utility patent generally runs twenty years from its earliest effective non-provisional filing date, so a patent that will expire before launch poses no forward-looking risk—though it can still support a claim for past damages if you were already in the market. Conversely, pending applications that have not yet issued may grant later with claims that read on the product, and their claims can change dramatically before allowance. A thorough FTO therefore catalogs both issued patents and published applications, treating the latter as a watch-list rather than a verdict.

Step two: search the landscape

With scope defined, the next task is to find the patents that might matter—a phase that is at once essential and inherently imperfect. No search guarantees that every relevant patent surfaces. Patents use idiosyncratic terminology (a "smartphone" might be a "portable wireless communication apparatus"), claims can be drafted broadly enough to cover things their authors never pictured, and new patents issue every Tuesday. The goal is a search thorough enough to catch the significant risks while accepting that certainty is unattainable—and, crucially, documenting the effort so the diligence is provable later.

The raw material lives in patent databases. The USPTO's Patent Public Search and Patent Center offer free access to U.S. patents and published applications; the European Patent Office's Espacenet covers more than a hundred jurisdictions; WIPO's PATENTSCOPE reaches international PCT applications. Commercial platforms—Google Patents, Derwent Innovation, PatSnap, Questel Orbit—layer analytics, machine translation, and aggregated international coverage on top. The search strategy itself braids together two complementary methods. Classification searching uses the Cooperative Patent Classification (CPC) and International Patent Classification (IPC) schemes to find patents that examiners have filed in a technical area regardless of the words they use; for Northgate that means CPC subclasses covering payment architectures (G06Q 20/-), cryptographic mechanisms (H04L 9/-), and security arrangements (G06F 21/-). Text searching complements it by hunting specific terms, which requires thinking like a patent drafter—brainstorming synonyms and alternative phrasings ("mobile payment," "digital wallet," "transaction authentication," "biometric verification," "contactless," "NFC payment") and refining them with Boolean operators, proximity connectors, and truncation. Searching the portfolios of known competitors, the major card networks and handset makers, and the patent-assertion entities active in payments rounds out the picture, because assignee searches catch patents that classification and keyword searches miss.

Two disciplines run through the whole phase. First, search scope is a permanent tension between thoroughness and practicality: broad searches return unmanageable result sets, narrow ones miss things. Most practitioners iterate—a broad first pass maps the terrain, reviewing the hits surfaces new terminology and classifications, refined searches target the danger zones, and the cycle repeats until coverage feels reasonable. Second, the search strategy must be documented: the databases queried, the classifications and keywords used, the exact strings, the dates run, and the results. That record is not bureaucratic box-ticking. If the product is later accused of infringement, or if the analysis is offered to negate willfulness, the documentation is the evidentiary backbone of the good-faith story—and it is what lets a junior associate efficiently update the search a year later as the product and the field move. Searching for FTO also differs in aim from searching for validity, a point we return to below: an FTO search hunts for patents whose claims might cover your product, while a validity search hunts for prior art that might invalidate their patent. They look in overlapping places for opposite reasons.

Step three: triage to the patents that actually matter

A good search returns far more patents than are relevant, and the next job is ruthless filtering. Three successive sieves do most of the work. The first is technological relevance: many hits are plainly outside the product's technology despite tripping a keyword—a mechanical coin-counter that surfaced under a payments classification is irrelevant to a software app—and can be set aside in seconds. The second is claim scope: even a technologically relevant patent matters only if its claims—not its title, not its abstract, not its glossy description, but its claims—could read on the product, so a quick read of the representative independent claims tells you whether a patent survives to the next round. The third is legal status: expired, lapsed, disclaimed, or already-invalidated patents pose no forward risk and can be eliminated, while a patent subject to a terminal disclaimer (which ties its term and enforceability to a related patent) should be flagged for special handling.

What emerges is a prioritized list. The patents at the top are those whose claims plausibly cover the product, which are currently in force, and which are owned by an entity with a track record—or a business model—of enforcement. Pending applications get their own bucket: because their claims may shift before issuance, they are monitored rather than adjudicated. For Northgate, the search might surface several hundred patents, the triage might cut that to a couple dozen worth a careful claim read, and the deep, opinion-grade analysis described in the rest of this article might land on a dozen or fewer. That funnel—hundreds to dozens to a handful—is the normal shape of a competent landscape study. Practical Law's guidance is blunt about why it works: experienced patent counsel who genuinely understand the client's technology can evaluate search results and pick out the high-risk patents quickly, reserving the expensive, granular analysis for the few that earn it.

Step four: learn to read claims

Patent claims are where FTO lives, because the claims—not the evocative "summary of the invention," not the figures, not the patent's marketing-flavored title—define the legal metes and bounds of what a patent covers. A patentee who fails to claim an aspect of an invention gets no rights to it, however lovingly the specification describes it. Reading and interpreting claims is therefore the core skill of FTO, and it has a logic worth internalizing.

A claim is a single sentence—however long, nested, and grammatically tortured—that defines the protected invention. Independent claims stand alone and recite every element; dependent claims incorporate an independent claim by reference and add limitations, so they are necessarily narrower. FTO usually concentrates on the independent claims, because if a product does not infringe the independent claim, it cannot infringe anything that depends from it (Monsanto Co. v. Syngenta Seeds, Inc., 503 F.3d 1352, 1359 (Fed. Cir. 2007)). And note the asymmetry that cuts the other way: a patent is infringed if even one of its claims is infringed (Grober v. Mako Products, 686 F.3d 1335, 1344 (Fed. Cir. 2012)). The building blocks of a claim are its elements (or "limitations"), and the cardinal principle is the all-elements rule: a product infringes only if it contains every element the claim recites. A claim to "a widget comprising A, B, and C" is not infringed by a product with only A and B, even if those two match perfectly. If the accused product is missing a single claimed element, there is no literal infringement (V-Formation, Inc. v. Benetton Group SpA, 401 F.3d 1307, 1312 (Fed. Cir. 2005)). This is the single most important idea in infringement analysis, and it is the one that most often supplies the escape route.

Claim language also carries specialized legal freight. Terms of degree—"substantially," "approximately"—introduce flexibility (and litigation). Transitional phrases are decisive: "comprising" is open-ended, so "comprising A, B, and C" is infringed by a product with A, B, C and D, whereas "consisting of" is closed, so the same claim with "consisting of" is not infringed by a product that adds D. And "means for [performing a function]" language presumptively invokes 35 U.S.C. § 112(f), which restricts the claim to the structure actually disclosed in the specification and its equivalents—though the Federal Circuit warned in Williamson v. Citrix Online, LLC, 792 F.3d 1339 (Fed. Cir. 2015) (en banc), that a "nonce" word like "module" or "mechanism" can trigger § 112(f) even without the word "means" if it recites function without sufficient structure. Sorting all of this out is claim construction, which is a question of law for the court rather than the jury under Markman v. Westview Instruments, Inc., 517 U.S. 370 (1996).

The methodology comes from Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) (en banc): claim terms carry their ordinary and customary meaning to a person of ordinary skill in the art at the time of the invention, read in light of the intrinsic evidence—the claims, the specification (the single best guide, which may define terms expressly or by implication), and the prosecution history (the applicant's recorded back-and-forth with the examiner, which can reveal arguments or amendments that narrowed the claim). Extrinsic evidence—dictionaries, treatises, expert testimony—is admissible but expressly less reliable, and a construction that leans on extrinsic evidence to contradict the intrinsic record is on thin ice (Phillips, 415 F.3d at 1318–19). For FTO, counsel must develop defensible constructions and, importantly, hunt for the narrow ones that help the client: a specification that repeatedly casts a single embodiment as "the invention" can limit claim scope (X2Y Attenuators, LLC v. Int'l Trade Comm'n, 757 F.3d 1358, 1362 (Fed. Cir. 2014)), and a clear, unequivocal disclaimer in the prosecution history can shrink a term below its ordinary meaning (Computer Docking Station Corp. v. Dell, Inc., 519 F.3d 1366, 1374–79 (Fed. Cir. 2008)). Counsel should also check whether the claims were already construed somewhere—an earlier district-court Markman order (which may bind the patentee), an ITC Section 337 investigation, or a PTAB proceeding—because a prior construction can powerfully shape the analysis (Tex. Instruments Inc. v. Cypress Semiconductor Corp., 90 F.3d 1558, 1569 (Fed. Cir. 1996)). And because courts construe claims to preserve validity where possible (Phillips, 415 F.3d at 1327), construction and the validity analysis in step six are intertwined from the start.

Step five: assess infringement—literally and by equivalents

With the claims construed, the analysis lays the product against each claim element. Direct infringement under 35 U.S.C. § 271(a) occurs when someone, without authorization, makes, uses, offers to sell, sells, or imports the patented invention in the United States, and the core FTO question is whether Northgate's app would do any of those things as to any claim of any in-force patent. (Counsel should also keep the indirect and cross-border theories in view—inducement under § 271(b), contributory infringement under § 271(c), and the export/import provisions of § 271(f) and (g)—because a payments platform that ships components or SDKs abroad, or imports goods made by a patented process, can trip them.)

Literal infringement exists when the product meets every claim element exactly as construed. The analysis is mechanical in form: does the app have element A? Element B? Element C? If every element is present, the product literally infringes that claim; if even one is absent or materially different, it does not. This is where the all-elements rule earns its keep, and where a single distinguishing feature clears a patent. (For a means-plus-function element, literal infringement additionally requires that the accused structure be identical or insubstantially different from the structure the specification discloses—Kemco Sales, Inc. v. Control Papers Co., 208 F.3d 1352, 1364 (Fed. Cir. 2000).)

But literal analysis is not the end, because of the doctrine of equivalents, which extends infringement to a product that falls outside the literal claim language yet is equivalent to the claimed invention. The Supreme Court preserved the doctrine in Warner-Jenkinson Co. v. Hilton Davis Chemical Co., 520 U.S. 17 (1997), building on Graver Tank & Mfg. Co. v. Linde Air Products Co., 339 U.S. 605 (1950), and framing equivalence through the familiar "function-way-result" test: an element is equivalent if it performs substantially the same function in substantially the same way to achieve substantially the same result—or, put differently, if the differences between the claimed element and the accused element are insubstantial. The doctrine exists to stop an infringer from escaping through a trivial change. But it is fenced by powerful limiting doctrines. Equivalence is assessed element by element, never invention-as-a-whole, and cannot be used to read a required element entirely out of a claim (DePuy Spine, Inc. v. Medtronic Sofamor Danek, Inc., 469 F.3d 1005, 1016–17 (Fed. Cir. 2006)). The dedication rule bars claiming as an equivalent any subject matter the specification disclosed but failed to claim. And above all, prosecution history estoppel—addressed in Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., 535 U.S. 722 (2002)—bars a patentee from recapturing through equivalents the territory it surrendered by a narrowing amendment made for a reason related to patentability, with a presumption that the patentee gave up everything between the original and amended claim language. For Northgate, this means two reciprocal questions on every close call: even where the app dodges a claim's literal words, does it nonetheless infringe by equivalents—and, conversely, has the patent's own prosecution history foreclosed the equivalents argument the patentee would otherwise make? That second question is why FTO requires pulling the file history, not just reading the issued claim.

The output of this step is not a binary yes/no but a calibrated risk characterization for each relevant claim: high where the product clearly or very likely infringes; moderate where infringement is plausible and there are real arguments both ways; low where infringement looks unlikely but cannot be ruled out; and none where the product does not infringe under any reasonable theory.

A worked example: Northgate's app against a real-looking claim (hypothetical)

To make this concrete, suppose the triage surfaced a patent—owned by a payments company with a history of enforcement—whose independent claim 1 reads, in simplified form: "A method for authorizing a mobile transaction comprising: (a) capturing a biometric identifier at a mobile device; (b) generating, on the device, a cryptographic token from the biometric identifier; (c) transmitting the token to a remote authorization server over an encrypted channel; and (d) receiving an authorization message in response." Northgate's counsel builds a claim chart—a two-column table lining up each claim element against what the app actually does. The claim chart is the workhorse document of FTO; everything else is commentary on it.

Element (a): the app captures the user's face at the phone—present. Element (c): it transmits over an encrypted channel to a remote server—present. Element (d): it receives an authorization message back—present. Three of four elements read directly on the app, which is exactly the kind of overlap that makes a patent worth analyzing in depth rather than dismissing. The whole case turns on element (b). The claim requires generating a cryptographic token from the biometric identifier—deriving the token mathematically from the faceprint itself. But Northgate's architecture works differently: the face match merely unlocks a credential already provisioned in the phone's secure element, and the token is generated from that stored credential, not from the biometric data, which never leaves the enclave and is never an input to the token. Under the all-elements rule, if the app lacks element (b) as properly construed, it does not literally infringe claim 1—no matter how perfectly it satisfies the other three.

That is not the end of the inquiry, because of the doctrine of equivalents. The patentee would argue that deriving a token from a stored credential that was itself unlocked by the biometric is equivalent to deriving it from the biometric—same function (authenticating the user), same general way, same result. Northgate would answer on two fronts. First, on the merits: the architectural separation is a substantial difference, not an insubstantial one, because deliberately keeping the biometric entirely out of the token-generation path is a meaningful security design choice with real consequences for how the system behaves under attack. Second, and more decisively, on estoppel: if during prosecution the applicant amended the claim to add "from the biometric identifier" in order to overcome a prior-art rejection, Festo presumes the applicant surrendered everything outside the amended language—including tokens derived from something other than the biometric—and bars the equivalents argument outright. This is the payoff of reading the file history. The patentee's own past arguments may have handed Northgate its defense, gift-wrapped.

The example also shows how a clearance analysis quietly becomes a design directive. Suppose, instead, that Northgate's original architecture had derived the token straight from the faceprint, squarely inside element (b). The FTO analysis would flag a high infringement risk—and the very same analysis would point at the fix: re-architect so the biometric only unlocks a separately stored credential, moving the product out of the literal claim and, if the change is substantial and prosecution history cooperates, out of the equivalents range too. Caught before the design freeze, that is a modest engineering ticket. Caught in a complaint, it is a recall, an injunction motion, and a willfulness fight.

FTO versus patentability: why owning a patent is not enough

It is worth pausing on the distinction that trips up even sophisticated founders, because the worked example makes it vivid. Suppose Northgate has obtained its own patent on exactly the security architecture just described—the separation of biometric matching from token generation. That patent is real, valuable, and entirely beside the point for FTO. It tells the world that Northgate's particular contribution is novel and non-obvious enough to deserve protection; it says nothing about whether shipping Northgate's overall app requires using other people's patented inventions—the NFC link, the tokenization scheme, the bill-splitting interface, the fraud-scoring model. A product can be densely blanketed by its inventor's own patents and still infringe a dozen third-party patents on features the inventor did not invent. Patentability looks inward at what you created; FTO looks outward at what others own. Northgate's patent is a sword it can wield against imitators and a bargaining chip in a cross-license; it is no shield against an infringement suit. Treating "we have patents" as a clearance opinion is one of the most common—and most dangerous—errors a launching company makes.

Step six: evaluate validity

A patent that is invalid cannot be infringed, so assessing the validity of the patents that pose real infringement risk adds a second dimension to FTO. Even where Northgate's app would read on a patent's claims, a strong invalidity argument shrinks the practical risk that patent poses. Validity analysis is usually reserved for the patents that survive step five as high—or at least moderate—infringement risks, because it is expensive and, as we will see, a non-infringement position is generally worth more than an invalidity one.

The most common validity challenges turn on prior art. Under 35 U.S.C. § 102, a claim is anticipated—invalid for lack of novelty—if a single prior-art reference discloses every element. Under 35 U.S.C. § 103, a claim is obvious if the differences between it and the prior art would have been obvious to a person of ordinary skill at the time. The Supreme Court's decision in KSR International Co. v. Teleflex Inc., 550 U.S. 398 (2007), reinvigorated the obviousness defense by rejecting the rigid "teaching-suggestion-motivation" formula in favor of a flexible, common-sense approach: obviousness can be shown where a skilled artisan would have had reason to combine known elements to predictable effect, or would have seen the claimed invention as an obvious solution to a known problem. Searching for validity prior art differs from the FTO search in both target and source: instead of patents covering the product, you are hunting references that disclose or suggest the patent's own claimed invention, and the search reaches beyond patents into technical literature, conference papers, product manuals, and published standards. (Our companion guide on overcoming obviousness rejections under Section 103 walks the same doctrine from the patent applicant's side of the table.)

Other validity attacks turn on the statute's disclosure requirements. A claim may be invalid for indefiniteness under § 112(b) if it fails to inform skilled artisans of its scope with reasonable certainty, the standard set in Nautilus, Inc. v. Biosig Instruments, Inc., 572 U.S. 898 (2014)—a real opening against the vague, functional claims that pervade software. A claim may fail the written-description or enablement requirements of § 112(a). And subject-matter eligibility under § 101 has become a potent weapon against software and business-method patents since Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014), which held that an abstract idea implemented on generic computer hardware is not patent-eligible—a doctrine of obvious relevance to a payments app, and one we treat at length in patent eligibility after Alice. For Northgate, Alice cuts both ways: it is a defense against many of the broad software patents the app might be accused of infringing, but it is also the reason those same patents were often drafted with functional, hard-to-pin-down claims that make construction unpredictable. Counsel may also note, but rarely rely on, unenforceability through inequitable conduct: because that defense turns on the patentee's intent to deceive the USPTO, it is usually unknowable without discovery, and an FTO analyst cannot responsibly conclude a patent is unenforceable on that ground from the public record alone. (We unpack that "atomic bomb" of patent law in inequitable conduct in patent prosecution.)

One caution governs the entire validity inquiry, and it must never be lost on a business reader. Invalidity is ultimately decided by a court or by the USPTO through proceedings like inter partes review (IPR)—not by the FTO analyst's say-so. Until a patent is actually canceled, it stays presumptively valid under 35 U.S.C. § 282(a), and an accused infringer who wants to prove invalidity in court bears a heavy burden: clear and convincing evidence (Apotex USA, Inc. v. Merck & Co., 254 F.3d 1031, 1037 (Fed. Cir. 2001)). That asymmetry—the patentee proves infringement by a mere preponderance while the challenger must clear the higher invalidity bar—is precisely why a clean non-infringement position is usually worth more than a strong invalidity argument. A good validity argument reduces risk; it does not erase it.

Confirming the patent is even alive: status, term, and the calendar

Before any of the substantive analysis is worth doing, counsel confirms the patent is actually in force and pins down its expiration date—and this is more involved than reading the issue date and adding twenty years. Several wrinkles can move the date in either direction. Patent term adjustment under 35 U.S.C. § 154(b) can extend the term to compensate for USPTO delays during prosecution, sometimes by years. Patent term extension under § 156 can extend it further for products that endured regulatory review (the classic case is a drug awaiting FDA approval). Cutting the other way, a patent can lapse early because the owner failed to pay a maintenance fee—though a lapsed patent can sometimes be revived under 37 C.F.R. § 1.378, so a recent lapse is not a guarantee of freedom. And a terminal disclaimer under § 253 and 37 C.F.R. § 1.321 can shorten the term and render the patent unenforceable if it is no longer commonly owned with the patent to which its term was tied. A practitioner who skips this step risks building an elaborate non-infringement opinion on a patent that expired eighteen months ago—or, worse, clearing a patent that turns out to have a longer life than the calendar suggested.

Step seven: willfulness and the role of opinions

The willfulness doctrine is the reason FTO analysis is not merely prudent but legally protective. Under § 284, a court may, in its discretion, increase damages up to three times the actual award when the infringer's misconduct is egregious, and § 285 lets the court award the prevailing party its attorneys' fees in an exceptional case. The prospect of trebled damages and a fee award is what makes clearing a product—and, in the right cases, paying for a formal written opinion—worth the cost.

The law here has shifted decisively in the accused infringer's favor on procedure but raised the stakes on substance. In Halo Electronics, Inc. v. Pulse Electronics, Inc., 579 U.S. 93 (2016), the Supreme Court swept away the rigid two-part test the Federal Circuit had imposed in In re Seagate Technology, LLC, 497 F.3d 1360 (Fed. Cir. 2007), and restored district courts' discretion to enhance damages in egregious cases. Halo relocated the inquiry to the infringer's subjective state of mind at the time of the challenged conduct—whether the company knew of the patent and knew, or was willfully blind to the fact, that what it was doing infringed. That makes the company's knowledge and good faith the whole ballgame, which is exactly where FTO and opinions of counsel come in: a timely FTO opinion concluding non-infringement or invalidity is evidence that, at the relevant time, the company genuinely believed it was in the clear (Omega Patents, LLC v. CalAmp Corp., 920 F.3d 1337, 1352–53 (Fed. Cir. 2019)).

An opinion of counsel—a formal written analysis concluding that the product does not infringe, or that the patent is invalid—has long served as evidence of good faith negating willfulness: if a company reasonably relied on a competent opinion, that reliance demonstrates the absence of the culpable mind that willfulness now requires. Critically, a company is not penalized merely for declining to obtain one. Congress codified that protection in 35 U.S.C. § 298: the failure to obtain advice of counsel, or to present such advice at trial, may not be used to prove willful infringement or inducement. There is no affirmative duty to get an opinion, and no adverse inference for skipping it. But a favorable, well-reasoned opinion remains valuable defensive evidence, and the catch is that a bad opinion can be worse than none. A superficial opinion, or one resting on aggressive claim constructions a court would never accept, reads like a company shopping for cover rather than seeking honest advice—and it invites a devastating cross-examination. A useful opinion construes claims reasonably rather than in the client's favor, reaches conclusions the analysis actually supports, is dated before the client engages in the potentially infringing activity (Critikon, Inc. v. Becton Dickinson Vascular Access, Inc., 120 F.3d 1253, 1259–60 (Fed. Cir. 1997)), and—because timing is about substance, not just paper—reflects analysis the client could realistically have acted on. Even an opinion obtained after a complaint arrives can help, if the company had no earlier notice of the patent (Vulcan Eng'g Co. v. Fata Aluminium, Inc., 278 F.3d 1366, 1379 (Fed. Cir. 2002)).

Several practical choices shape an opinion's value:

Non-infringement versus invalidity. Where counsel concludes the product does not infringe, a non-infringement opinion alone usually suffices to defeat willfulness; counsel need not also opine on invalidity (Finisar Corp. v. DirecTV Group, Inc., 523 F.3d 1323, 1339 (Fed. Cir. 2008)). A non-infringement opinion is generally preferable for two reasons. First, the burdens of proof favor it: the patentee must prove infringement by a preponderance, while the infringer must prove invalidity by clear and convincing evidence, so a non-infringement position rests on firmer ground. Second—and this is easy to overlook—an invalidity opinion does not defend against an inducement claim. Inducement under § 271(b) requires that the defendant knew the induced acts constituted infringement (Global-Tech Appliances, Inc. v. SEB S.A., 563 U.S. 754, 765–66 (2011)), and the Supreme Court held in Commil USA, LLC v. Cisco Systems, Inc., 575 U.S. 632 (2015), that a good-faith belief in invalidity is no defense to inducement, because validity and infringement are separate questions. A good-faith belief in non-infringement, by contrast, negates the very knowledge inducement requires. For a platform like Northgate's that supplies an app to millions of users (who arguably perform the patented "method" when they pay), the inducement exposure is real, and that alone may make a non-infringement opinion the right tool.

Who writes it. The opinion should be authored by a U.S. patent attorney with a background in the relevant technology. Beyond that, either in-house or outside counsel can write it; a competent in-house opinion can still establish good faith (Cohesive Techs., Inc. v. Waters Corp., 526 F. Supp. 2d 84, 106 (D. Mass. 2007)). But for high-risk patents, outside counsel is often preferable because a court may view outside counsel as more objective (Gaus v. Conair Corp., 2003 WL 223859, at *17 (S.D.N.Y. 2003)), and the opinion's author should ideally not be the company's trial counsel—both to keep the witness credible and to avoid an unnecessarily broad privilege waiver.

Privilege, and the waiver trap. This is where opinions get dangerous if mishandled. An FTO opinion prepared by counsel for legal advice is protected by attorney-client privilege and work-product doctrine—but the moment a company relies on the opinion as a willfulness defense in litigation, it waives privilege over all attorney-client communications on that subject matter (In re EchoStar Communications Corp., 448 F.3d 1294, 1298–99 (Fed. Cir. 2006)), and the waiver can sweep in communications created before and after the opinion's date. Because of that, sophisticated counsel often prepare separate opinions for non-infringement and invalidity, and separate opinions for different patents or products, so the client can choose to rely on one without involuntarily exposing the others (courts split on how far that strategy holds). Counsel should also warn the client not to share the opinion with third parties—in an acquisition or licensing diligence, for instance—without a common-interest agreement, lest the sharing itself waive the privilege. The same confidentiality discipline that governs other protected legal analysis applies here; we describe it in the context of sensitive technical materials in trade secrets in the age of remote work and cloud computing.

Whether to obtain an opinion at all is a cost-benefit judgment driven by the size of the investment, the expected revenue and lifespan of the product, the competitiveness of the market, and the company's risk tolerance. For a high-risk patent held by a likely enforcer guarding a flagship launch, a formal opinion is usually worth it. For a low-risk patent or an unlikely enforcer, the underlying FTO analysis may itself be enough to show good faith.

Step eight: design-arounds, licensing, and the other roads to "yes"

FTO is not only an assessment exercise. When it surfaces a genuine risk, it becomes a problem-solving exercise, and there are several roads to freedom to operate beyond simply abandoning the technology.

A design-around modifies the product to avoid infringement, and it starts by identifying which claim element creates the risk. Because of the all-elements rule, eliminating or substantially changing any single element of a claim avoids literal infringement of that claim—so if a problematic claim requires A, B, C, and D and Northgate's app has all four, changing even one may clear it. Some design-arounds are cheap (swap an alternative method step that reaches the same result); others are painful, because the claimed feature is central to the product's value or is simply the best engineering solution—and because designing around one patent can create exposure under another. The doctrine of equivalents complicates the exercise: a change that escapes the literal language may still infringe by equivalents if it does substantially the same thing in substantially the same way, so the design-around must be substantial enough to exit the equivalents range, not a cosmetic relabeling a court will see through. Documenting the design-around reasoning is worthwhile, because evidence that a product was deliberately engineered around a patent, on a reasonable reading of its claims, supports both non-infringement and the absence of willfulness. The cleaner play, where the development timeline allows, is to feed FTO results back to engineering before the design freeze, so risky features can be re-architected rather than ripped out.

Licensing offers a transactional path: if the patent owner grants permission, infringement liability disappears, which is why understanding ownership is part of FTO. Who holds the risky patents—a direct competitor likely to enforce, a university whose technology-transfer office licenses as a matter of institutional mission, a patent-assertion entity that exists to collect royalties, or a company with which Northgate already has a commercial relationship? The owner's identity drives both the likelihood of enforcement and the prospects for a deal, and existing cross-licenses may already cover the new product, so reviewing the company's own license portfolio is part of the workflow. In some cases the owner may simply sell the patent—attractive when the patent is valuable to Northgate but peripheral to the owner, or when the owner is a distressed seller—converting a liability into an asset. For the mechanics of valuing and structuring a deal once that path is chosen, see how to license your patent: from valuation to term sheet.

The patent-assertion entity (PAE) deserves special mention, because payments is favored hunting ground and PAEs distort the risk calculus in ways that have little to do with technical merit. A PAE typically makes no product, which means Northgate cannot counter-sue it for infringing Northgate's patents—the cross-licensing leverage that disciplines disputes between operating companies simply is not on the table. PAEs also tend to price their demands deliberately below the cost of defense, so paying to settle can look rational even against a patent the company believes is invalid or not infringed. FTO should therefore flag PAE-held patents not just by their technical risk but by the realistic economics of being targeted, and it should feed an early strategic decision: design around proactively, challenge a weak patent through IPR before any demand arrives, or budget for the nuisance-value settlements a consumer-facing payments product may simply attract. Knowing who holds a patent is, in this sense, as important to the risk assessment as knowing what it claims.

Step nine: document, then communicate risk honestly

Two final steps convert analysis into business value: documenting it, and communicating it.

Documentation serves several ends at once. It creates a record that can be efficiently updated; it transfers knowledge as personnel turn over; it evidences good-faith clearance if litigation comes; and it anchors the conversation with decision-makers. A complete FTO report describes the scope analyzed (the product specification, the geography, the search date), the search methodology (databases, classifications, keywords, strings), the patents identified and the triage that narrowed them, and—for each patent analyzed in depth—the claim charts mapping elements to product features, the claim-construction analysis of disputed terms, the infringement risk assessment with reasoning, any validity issues, and an overall risk characterization. It closes with an executive summary that translates the technical analysis into plain-English guidance. And, as step seven explained, the whole package should be handled with privilege in mind: prepared by or under counsel for the purpose of legal advice, and not casually circulated.

Communication is where FTO meets the boardroom, and the governing discipline is honesty about uncertainty. Risk should be calibrated to the audience: executives need a high-level characterization—green light, yellow caution, red stop—with enough context to understand it; product and engineering teams need the detail on specific patents and design-around options; legal teams need the full analysis. Risk should not be dressed in false precision. Patent infringement risk depends on how a court would construe contested terms, how a factfinder would assess infringement and validity, and whether the owner would even sue—none of which collapses into a clean probability. A qualitative category with explanatory context ("high risk: the app reads on three of four elements; the fourth is the only thing standing between us and literal infringement, and the patentee owns it") communicates the uncertainty far more honestly than a spuriously specific "30% likelihood." Finally, every identified risk should travel with its mitigation options—design-around, license, IPR challenge, accept-and-reserve, acquire the patent, or wait for expiration—each tagged with its costs, timeline, and residual risk, so the business can make an informed call. Because the right answer is contextual: a high-risk patent may be perfectly acceptable when projected revenue dwarfs the worst-case exposure, while a moderate-risk patent may be unacceptable when the owner is a known aggressive enforcer. FTO supplies the information; the business makes the decision.

FTO is not a one-time event

A clearance that was accurate at launch can go stale, because the landscape moves. New patents issue every week, including continuation applications that competitors sometimes draft specifically to read on a known product—a practice sometimes called "patent tracking"—so standing search alerts on relevant classifications, keywords, and competitor assignees should run on an ongoing basis. Product changes alter the analysis too: a new feature implicates new patents, and a cost-driven design change can shift the infringement picture for a patent already cleared. And outside developments matter—a court construing a patent's claims, a PTAB or court invalidity ruling, an acquisition that puts a sleepy patent into an aggressive owner's hands, or a license deal that changes the field. Building FTO monitoring into the product lifecycle—an annual review, or a review tied to each major release—keeps the analysis current for products that carry meaningful risk.

Different technologies, different challenges

The methodology above is general, but three fields stress it in distinctive ways.

Software and business-method patents remain plentiful despite Alice, and they carry broad, functional claims that describe what software does rather than how it does it—which makes infringement turn on unpredictable construction outcomes. The sheer volume of software patents and the abstractness of software functionality make comprehensive searching genuinely hard, which is Northgate's daily reality and the reason its FTO must be both broad and iterative. The flip side is that the same patents are often the most vulnerable to a § 101 or § 112(b) attack. For the broader strategy of layering patents, copyrights, trade secrets, and contracts around a mobile product, see protecting your mobile app: a comprehensive IP strategy guide.

Pharmaceutical and life-sciences patents raise different issues, where regulatory exclusivities interlock with patents. The FDA's Orange Book (small-molecule drugs) and Purple Book (biologics) list patents that frame the generic- and biosimilar-entry landscape, and the Hatch-Waxman framework gives infringement under § 271(e) a specialized choreography. Chemical and biological structures make claim mapping more determinate than in software—you can often tell whether a molecule reads on a claim—but the enormous stakes drive aggressive enforcement, and the "insubstantial differences" branch of the doctrine of equivalents tends to do more work than function-way-result in the chemical arts.

Standard-essential patents create their own puzzle: implementing a technical standard necessarily practices the patents essential to it, so a product that complies with a communication or security standard may infringe by definition. Standards bodies usually require members to license essential patents on fair, reasonable, and non-discriminatory (FRAND) terms, but defining and enforcing FRAND has generated enormous global litigation. Any product implementing a standard—as a payments app implementing connectivity and security standards may—must analyze the standards-essential landscape, a subject we examine in standard-essential patents and FRAND licensing in 5G and IoT.

There is also an instructive parallel outside patents entirely. Trademark law runs the same "clear before you launch" playbook: a brand owner searches for conflicting marks, weighs likelihood of confusion, and may obtain a clearance opinion to blunt a later willfulness claim. The doctrines differ, but the risk-management instinct is identical, and a company launching a new product usually needs both clearances. Our guide on how to conduct a comprehensive trademark clearance search is the trademark-side companion to this one, and the shield of good faith explains how clearance and opinions cut off willfulness in that context.

Frequently asked questions

Is freedom to operate the same as patentability? No, and conflating the two is the classic mistake. Patentability asks whether your invention is new and non-obvious enough to earn a patent; FTO asks whether selling your product would infringe someone else's patent. You can have one without the other. Holding your own patents proves nothing about your freedom to operate.

Do we legally have to get an FTO opinion before launching? No. There is no affirmative duty to obtain an opinion of counsel, and 35 U.S.C. § 298 forbids using the absence of an opinion to prove willful infringement or inducement. The reason to get one is strategic, not mandatory: a competent, timely opinion is strong evidence of the good faith that defeats a willfulness claim under Halo, and that protection is most valuable for a large, high-visibility launch.

Does a strong invalidity argument mean we're safe? Not on its own. A patent stays presumptively valid under § 282 until a court or the USPTO actually cancels it, and an accused infringer must prove invalidity by clear and convincing evidence—a demanding standard. A strong invalidity position reduces risk and can support an IPR challenge or a settlement, but it does not eliminate exposure. And remember that a good-faith belief in invalidity is no defense to an inducement claim, which is why a non-infringement position is usually worth more.

What is a claim chart and why does everyone keep mentioning it? A claim chart is a two-column table that places each element ("limitation") of a patent claim beside the corresponding feature of your product, with an assessment of whether the feature meets the element. It is the central work product of an FTO analysis because infringement is decided element by element under the all-elements rule. If your product is missing even one element, there is no literal infringement of that claim—and the chart is where you find the missing element.

A patent-assertion entity sent us a demand letter. Is it different from a competitor suit? In practice, yes. A PAE makes no product, so you cannot counter-sue it for infringing your patents, and PAEs often price demands just below your cost of defense, which makes paying look rational even against a weak patent. FTO should flag PAE-held patents for the economics of being targeted, not just their technical merit, and you should weigh proactive design-arounds or an early IPR challenge against simply budgeting for nuisance settlements.

How often should we redo our FTO? Treat it as a living process for any product with meaningful risk. New patents and continuation applications issue constantly, your own product changes, and court or PTAB rulings move the field. Standing search alerts plus a scheduled review—annually, or tied to major releases—keeps the analysis from going stale.

Conclusion

Freedom-to-operate analysis is risk management, not a guarantee of safety. No search finds every patent; no claim construction is certain; no infringement or validity assessment is immune from revision as the facts develop. What FTO delivers is informed decision-making—an understanding of the patent landscape good enough to assess the risks intelligently and allocate resources sensibly. The investment should be proportionate to the stakes: a major launch like Northgate's, with years of work and millions of dollars riding on it, warrants comprehensive searching, careful claim charts, and formal opinions for the serious risks, while a minor variation on an already-cleared product may need only a focused look at what changed.

The discipline ultimately serves the business by finding the landmines before they detonate. The cost of analysis is modest beside the cost of an injunction that pulls a launched product from the shelves, or a trebled-damages judgment for willful infringement, or the slow bleed of nuisance settlements to a patent troll who saw you coming. By investing in a thorough search, careful claim analysis, honest risk assessment, and clear communication, a company can bring its product to market with appropriate—if never absolute—confidence in its freedom to operate. For Northgate, that confidence is what lets it spend launch week worrying about servers and customers instead of a lawsuit it could have seen coming.

For help conducting a freedom-to-operate analysis or preparing opinions for a product launch, contact our intellectual property and technology practice or our patent team.

Related articles

Selected authorities

35 U.S.C. §§ 101, 102, 103, 112, 154, 156, 253, 271, 282, 283, 284, 285, 298. 37 C.F.R. §§ 1.321, 1.378. Markman v. Westview Instruments, Inc., 517 U.S. 370 (1996); Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) (en banc); Graver Tank & Mfg. Co. v. Linde Air Products Co., 339 U.S. 605 (1950); Warner-Jenkinson Co. v. Hilton Davis Chemical Co., 520 U.S. 17 (1997); Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., 535 U.S. 722 (2002); DePuy Spine, Inc. v. Medtronic Sofamor Danek, Inc., 469 F.3d 1005 (Fed. Cir. 2006); KSR International Co. v. Teleflex Inc., 550 U.S. 398 (2007); Nautilus, Inc. v. Biosig Instruments, Inc., 572 U.S. 898 (2014); Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014); Williamson v. Citrix Online, LLC, 792 F.3d 1339 (Fed. Cir. 2015) (en banc); Halo Electronics, Inc. v. Pulse Electronics, Inc., 579 U.S. 93 (2016); In re Seagate Technology, LLC, 497 F.3d 1360 (Fed. Cir. 2007); Omega Patents, LLC v. CalAmp Corp., 920 F.3d 1337 (Fed. Cir. 2019); Global-Tech Appliances, Inc. v. SEB S.A., 563 U.S. 754 (2011); Commil USA, LLC v. Cisco Systems, Inc., 575 U.S. 632 (2015); Finisar Corp. v. DirecTV Group, Inc., 523 F.3d 1323 (Fed. Cir. 2008); In re EchoStar Communications Corp., 448 F.3d 1294 (Fed. Cir. 2006); Critikon, Inc. v. Becton Dickinson Vascular Access, Inc., 120 F.3d 1253 (Fed. Cir. 1997); Vulcan Eng'g Co. v. Fata Aluminium, Inc., 278 F.3d 1366 (Fed. Cir. 2002); Cohesive Techs., Inc. v. Waters Corp., 526 F. Supp. 2d 84 (D. Mass. 2007); Kemco Sales, Inc. v. Control Papers Co., 208 F.3d 1352 (Fed. Cir. 2000); Monsanto Co. v. Syngenta Seeds, Inc., 503 F.3d 1352 (Fed. Cir. 2007); V-Formation, Inc. v. Benetton Group SpA, 401 F.3d 1307 (Fed. Cir. 2005); Grober v. Mako Products, 686 F.3d 1335 (Fed. Cir. 2012); X2Y Attenuators, LLC v. Int'l Trade Comm'n, 757 F.3d 1358 (Fed. Cir. 2014); Apotex USA, Inc. v. Merck & Co., 254 F.3d 1031 (Fed. Cir. 2001).

This guide is for general informational purposes only and does not constitute legal advice, nor does it create an attorney-client relationship. Patent law varies by jurisdiction and continues to develop; the discussion here may not reflect the most recent developments. Freedom-to-operate analysis is fact-specific—consult qualified patent counsel about your particular product and markets before launch.