Document type: Guide Practice area: Intellectual Property — Patents Jurisdiction: United States Last reviewed: 5 September 2026

Before you start: know which negotiation you are in

Patent licences arise from three quite different situations, and treating them the same is the first mistake.

The technology transfer. A university, a research institute, or a company with a technology it does not intend to commercialize licenses to someone who will. Both parties want the licence to succeed. The negotiation is about risk allocation and upside sharing, and the leverage is roughly balanced.

The commercial cross-licence. Two operating companies with overlapping portfolios agree to stop worrying about each other. The negotiation is about relative portfolio value, scope, and duration, and it usually resolves into a balancing payment. Neither party is a supplicant.

The assertion licence. A patent owner has accused a company of infringement, and a licence is the way the accusation ends. The negotiation is a settlement, conducted in the shadow of litigation cost, and the terms will be read later by every other defendant.

The document that comes out of each looks superficially similar. The provisions that matter, the diligence you run, and the concessions worth making differ substantially. Establish which one you are in before you draft.

Step one: identify what is actually being licensed

Ask for the patent list, and then check it.

Patent numbers, not families. "The Licensor's patents relating to X" is not a grant; it is a dispute. Ask for a schedule of issued patents and pending applications by number, with jurisdiction, filing date, priority date, and expected expiration.

Ownership. Pull assignment records. Chain-of-title defects are common, especially where inventors moved between employers, where a startup was acquired, or where a university's inventor assignment was never executed. A licence from someone who does not own the patent is worth what it cost to draft.

Co-owners. Under U.S. law, each co-owner of a patent may practise and license it without the others' consent and without accounting to them. A licence from one co-owner therefore gives you rights, but it does not give you exclusivity and does not stop the other co-owner from licensing your competitor. If exclusivity matters, you need every co-owner.

Encumbrances. Prior licences, options, security interests, government rights under Bayh–Dole for federally funded inventions, standards commitments, and consent decrees. Ask specifically about each.

Term. Expiration dates, including terminal disclaimers and patent term adjustment. A portfolio whose key patent expires in three years is worth less than one whose key patent expires in twelve, and the royalty schedule should reflect it.

Validity and enforceability posture. Has the patent survived an inter partes review? Is one pending? Is there prosecution history that will narrow the claims? Ask, and read the file history for the claims you care about.

Step two: understand what you need to do

The grant must match the business, and the business is often described imprecisely by the people who run it.

Sit down with the engineers and the commercial team and answer these questions concretely:

  • What products will practise the patents, now and in the foreseeable roadmap?
  • Will you manufacture, or will a contract manufacturer? If a third party will manufacture, you need have-made rights.
  • Will you sell components to others who will complete the product? If so, you need to think about whether your customers need rights.
  • Will you sell through distributors and resellers? Do they need rights, or does your authorized sale cover them?
  • What territories? Where will manufacture occur, and where will sale occur? These may differ.
  • What fields? Are you willing to accept a field limit in exchange for a lower rate, and if so, is the field defined in a way that will not strangle you in three years?
  • Do you need to sublicense? To affiliates? To joint venture partners? To customers?
  • Do you need the right to modify and improve, and who owns the improvements?

Answer these before the first call. A grant clause negotiated from a clear specification takes an hour. One negotiated while the licensee's own team argues about the roadmap takes a month.

Step three: run the freedom-to-operate reality check

A licence solves the problem it addresses. It does not solve the problem you have not looked at.

Before paying for a licence, know whether it actually clears your path. If the licensor holds three of the seven patents that read on your product, a licence to those three has value but does not let you launch. Conversely, if your product does not infringe at all, the correct answer to a licensing demand may be a well-supported non-infringement position rather than a cheque.

Run, or refresh, a freedom-to-operate analysis scoped to the product and the claims at issue. Have the technical team map each asserted claim element to the accused product. Where an element is absent, say so early and specifically — a licensor confronted with a credible non-infringement chart negotiates differently.

And be honest internally about the alternative. Design-around cost, delay, and performance impact are the real benchmark for what a licence is worth. If a design-around costs $400,000 and eighteen months, a $3 million licence demand has an answer.

Step four: draft the grant

The grant clause is where the value sits. Everything else is machinery.

Enumerate the rights. Make, have made, use, offer for sale, sell, and import are separate rights. Grant the ones you need and understand what you have not been granted.

Exclusivity. Decide among exclusive, sole (licensor may practise but will not license others), and non-exclusive. Exclusivity costs more and carries diligence obligations. Sole licences are underused and often the right answer where the licensor wants to retain research use.

Field. If the grant is field-limited, define the field by reference to something objective — end product, technical specification, regulatory classification, customer type. Then apply the definition to your roadmap and see what falls outside it. Add a mechanism to expand the field: an option, a right of first negotiation, or a stated rate for additional fields.

Territory. Manufacture and sale territories may differ. Note that patents are national rights; a licence to a U.S. patent is territorially limited by the patent itself.

Have-made. Include it if you use contract manufacturers. Expect a limitation to products made to your specifications, for your account, and sold by or for you. Accept it.

Sublicensing. State whether it is permitted, whether consent is required, whether consent may be withheld unreasonably, whether sublicensees must accept the field limitation, what royalties flow through, and — the provision everyone forgets — whether sublicences survive termination of the head licence. As a sublicensee, insist that they do, with the sublicensee stepping into a direct relationship with the licensor on the same terms.

Affiliates. Define affiliate, state whether the grant extends to affiliates, and address what happens when an entity ceases to be an affiliate — a real problem in a carve-out or divestiture.

Step five: negotiate the money

Base. The royalty base is negotiated more than the rate and matters at least as much. Options:

  • Per-unit. Simple, predictable, and immune to accounting disputes. Best where units are countable and prices stable.
  • Percentage of net sales of the licensed product. The most common structure. The fight is over what "licensed product" means and what "net sales" deducts.
  • Percentage of the smallest saleable unit. Where the patent covers a component of a larger product, apportionment is the legally correct approach and the licensee should push for it.
  • Total sales base with a lower rate. Permissible where the licensee genuinely prefers it for administrative convenience — but a coerced total-sales base is misuse, so the record should show the licensee proposed or accepted it freely.

Net sales, defined. Deduct only: taxes actually paid, freight and insurance separately invoiced, trade discounts actually granted, and credits for returns. Exclude explicitly: overhead, marketing, cost of goods, bad debt, and any charge payable to the licensee or an affiliate.

Rate. Benchmark against comparable licences in the field, the profit margin on the product, the patent's contribution to that margin, the remaining term, and the strength of the patents. A rate that exceeds the licensee's margin is not a licence; it is a shutdown.

Structure. Lump sum, running royalty, or hybrid. Lump sums are clean and end the relationship's administrative burden; they also mean the licensor bears the risk that the product succeeds beyond expectations, and the licensee bears the risk that it fails. Running royalties align interests and create ongoing obligations. Hybrids — an up-front payment creditable against running royalties — are the usual compromise.

Minimums. A licensor granting exclusivity needs annual minimums or the exclusive field can be warehoused. A licensee accepting minimums should insist that the consequence of a shortfall is conversion to non-exclusive rather than termination, and should negotiate the minimum against a realistic forecast rather than an optimistic one.

Milestones. Common in technology transfer: payments on first regulatory filing, first approval, first commercial sale, and cumulative sales thresholds. Define each milestone so that its occurrence is objectively determinable.

Step-downs. Rates should decline as patents expire, and must not extend past the last-to-expire licensed patent without allocation to non-patent consideration.

Stacking. If the licensee must take other licences to practise, negotiate a stacking provision reducing the royalty by a portion of third-party royalties, with a floor.

Step six: reporting and audit

Reports should be quarterly, due 45 days after quarter end, and should state units by product and territory, gross sales, each deduction itemized, net sales, the rate applied, and the amount due. A report that gives only a total is not a report.

The audit clause needs: an annual right, thirty days' notice, an independent accountant of the licensor's choosing, access to underlying records, a records retention period matching the objection period, and cost shifting at a stated variance. Without the cost shift, audits do not happen.

For a licensee, the provisions to negotiate are confidentiality of the audit (the accountant reports only the amount of any discrepancy, not the licensee's underlying business data), a limit of one audit per year, and a cap on how far back an audit may reach.

Step seven: improvements

Improvements clauses cause more post-signature litigation than any other provision except the field definition, because they are usually drafted vaguely and because both parties assume the vague language means what they want.

Decide four things and write them down:

  1. What is an improvement? Anything within the claims? Anything that would infringe? Anything relating to the licensed technology? The narrowest workable definition is something that cannot be practised without infringing the licensed patents.
  2. Who owns improvements each party makes? The default should be that each party owns what it invents.
  3. What rights does the other party get? A grant-back licence is common. It should be non-exclusive, and for a licensor a grant-back that is exclusive or that assigns ownership is aggressive and can raise misuse concerns.
  4. Are the licensor's future improvements included? A licensee usually wants improvements made during the term included automatically. A licensor usually wants to charge for them. The compromise is inclusion of improvements dominated by the licensed patents, plus an option on others.

Step eight: enforcement and standing

Decide who can sue and settle before you need to know.

A non-exclusive licensee never has standing. An exclusive licensee with all substantial rights may sue alone. An exclusive licensee with less may sue only with the patentee joined.

If enforcement matters to the licensee, the licence should provide: the licensee's right to enforce within its field; a first-right structure under which the licensor may enforce and, if it declines within a stated period, the licensee may; the licensor's agreement to be joined as a necessary party at the licensee's cost; control of the litigation; a settlement approval mechanism that protects the licensor's rights outside the field; and an allocation of recoveries after costs.

If enforcement does not matter to the licensee, the provision still should say what happens when the licensee discovers infringement — at minimum a notice obligation and a cooperation obligation.

Step nine: representations, indemnity, and the limits of both

Patent licensors, especially universities and especially in non-exclusive licences, will not warrant validity or non-infringement, and the market accepts that. What a licensee can reasonably get:

  • Ownership and authority. The licensor owns or controls the patents and has the right to grant.
  • No conflicting grants. No prior licence or encumbrance inconsistent with this one.
  • No knowledge of invalidity. Often qualified to actual knowledge of named individuals.
  • Disclosure of litigation and proceedings involving the patents.
  • Bayh–Dole disclosure where federal funding was involved, including march-in rights and the U.S. manufacturing preference.

What a licensee generally cannot get: a warranty that the patents are valid, that practising them will not infringe a third party's rights, or an indemnity for third-party infringement claims. The absence of the last is worth explaining to the business team, who often assume a licence clears the field.

Running the other way, licensors typically require the licensee to indemnify for product liability arising from licensed products, and to carry insurance naming the licensor as an additional insured. This is standard and reasonable.

Step ten: term, termination, and what survives

Term. Until the last-to-expire licensed patent, or a stated period, whichever the parties prefer. Be explicit about what happens to royalty obligations as patents expire.

Termination for breach. Notice and cure, thirty to sixty days for payment breaches, longer for others. A licensee should resist termination for a payment dispute raised in good faith.

Termination for convenience by the licensee. Reasonable in a running-royalty licence and usually granted on ninety days' notice.

Termination on validity challenge. Enforceable, and licensors want it. A licensee should try to limit it to challenges the licensee itself initiates, excluding a challenge by an acquirer or one raised defensively in litigation the licensor started.

Insolvency. Address the bankruptcy protections available to intellectual property licensees, and be aware of their limits. Note in particular that trademarks are treated differently from patents in the relevant statutory definition, which matters where the licence covers both.

Survival. Confidentiality, accrued payment obligations, audit rights for a period, indemnities, and the disposition of inventory on hand.

Sell-off. A licensee terminating or expiring should have a period to sell finished inventory, with royalties payable.

Step eleven: the provisions that decide what a licence is worth in five years

A licence signed today will be read, most likely, by people who were not in the room. Some provisions age well and some do not.

Most favoured licensee. A licensee may ask for a promise that if the licensor grants better terms to anyone else, the licensee gets them. Licensors resist, for good reason: an MFL provision constrains every future deal and can prevent the licensor from settling cheaply with a small infringer. Where an MFL is granted, it should be limited to substantially similar grants in the same field with comparable volumes, should exclude settlements of litigation, should require the licensee to take the entire other licence rather than cherry-pick its best terms, and should have a notice-and-election mechanism rather than automatic adjustment. An unqualified MFL clause is a landmine that detonates on a deal nobody is thinking about yet.

Escalation and inflation. A per-unit royalty fixed in nominal dollars declines in real terms over a fifteen-year patent life. Licensors should index; licensees should cap the index.

Currency. Where sales occur in multiple currencies, specify the conversion rate source, the conversion date, and who bears conversion cost. This is a small provision that becomes a large dispute in a volatile year.

Withholding tax. Cross-border royalty payments attract withholding. Decide who bears it, whether payments are grossed up, and whether the parties will cooperate on treaty relief certificates. Gross-up provisions are expensive and are frequently agreed without anyone modelling the cost.

Late payment. Interest at a stated rate, running from the due date. Modest, but it changes behaviour.

Notices. Include email as a valid channel and name a position rather than a person. Licences outlive the individuals named in them, and a notice sent to a departed employee's address is a notice that was never given.

Dispute resolution. Consider carving royalty and audit disputes out to expert determination by an accounting firm — faster and cheaper than litigation, and appropriate for a question that is arithmetic rather than legal. Keep infringement, validity, and scope questions for the courts or for arbitration, as the parties prefer.

Step twelve: run the exhaustion analysis before you sign

Both sides should do this, and in practice neither usually does.

For the licensor. Trace the licensed product through the chain. Who buys it? What do they do with it? Does the licensed product substantially embody any other claim you own or intend to assert? If a licensee's authorized sale will exhaust your system claims against the entire downstream industry, you need to know that before you set the rate — because you are selling far more than you think you are.

The specific questions:

  • Does the licensed article have any reasonable non-infringing use? If not, its sale exhausts the claims it embodies.
  • Are there method claims practised when the article is used as intended? Those are exhausted too.
  • Are there claims directed to different actors performing different steps, not embodied in the article? Those may be reserved — but only if the reservation is explicit and the claims are genuinely distinct.
  • Is the field limitation on the grant, or in a covenant? Only the first prevents exhaustion.

For the licensee. The mirror image. If you are buying a component from a licensed supplier, you may already have the rights you are being asked to pay for. A licensor demanding a licence for your use of a component it authorized its licensee to sell you is demanding payment for something it has already sold. Ask for the supplier's licence terms — suppliers are often willing to confirm the scope of their grant, and a single sentence confirming that the supplier's sale was authorized can end a licensing demand.

This analysis takes an afternoon and has, in more than one negotiation, converted a seven-figure demand into a letter.

Step thirteen: the negotiation itself

A few observations about how these conversations actually go.

Establish the technical facts early and separately. Whether a product infringes is a technical question. It is best resolved by engineers looking at claim charts, not by lawyers exchanging positions. Get the technical conversation done first; the commercial conversation is much shorter once both sides agree on what the product does.

Do not negotiate the rate before the base. A licensor that opens with "four percent" and a licensee that counters with "one percent" are arguing about a number neither has defined. Settle the base — what product, what deductions, what apportionment — and the rate often follows.

Trade scope for money. The most productive move in most patent licensing negotiations is to narrow the field in exchange for a lower rate, or to widen it in exchange for a higher one. Scope is the currency both sides value and neither has fully priced.

Bring the alternative into the room. Design-around cost, invalidity contentions, and non-infringement positions are legitimate parts of the discussion. Presented as analysis rather than as threat, they move a rate more than any argument about comparables.

Know when the relationship matters. In a technology transfer or a cross-licence, the parties will work together afterward. Winning a provision that poisons the relationship is a bad trade. In an assertion licence, there is no relationship and no reason to concede anything for goodwill.

Write the summary before you sign. One page, plain language: what was granted, in what field, in what territory, for how long, at what rate, with what obligations, terminable how. If the business team reads that page and says "wait, I thought we could also do X," you have found a problem while it is still free to fix.

Worked example one: the university licence

Dr. Amara Osei founded Tessellate Bio to commercialize an assay technology developed at a state university. The university's technology transfer office offers its standard exclusive licence.

The standard form has several provisions Amara's counsel, Rafael Duarte, negotiates.

Diligence. The form requires a first commercial sale within four years, with automatic termination on failure. Rafael converts automatic termination into a notice-and-cure structure with a right to extend on payment of an extension fee, and negotiates milestones tied to regulatory steps the company controls rather than calendar dates it cannot.

Field and reservation. The university reserves the right to practise for research and to grant non-exclusive research licences to other non-profits — standard and acceptable. It also initially reserves the right to license the technology for "diagnostic applications," which is precisely Tessellate's market. Rafael establishes that the reservation was boilerplate and gets it removed.

Bayh–Dole. The invention was federally funded. Rafael confirms the university filed its election to retain title and its confirmatory licence to the government, checks the U.S. manufacturing preference and its waiver process, and explains march-in rights to the board — which have never been exercised to grant a licence, but which investors will ask about.

Patent costs. The form makes Tessellate responsible for all past and future patent prosecution costs. Rafael accepts future costs (with a right to abandon jurisdictions Tessellate does not want, on notice) and negotiates past costs into a payment schedule.

Sublicensing. Tessellate needs to sublicense to a diagnostics partner. Rafael gets a sublicensing right subject to notice, a share of sublicensing income at a declining rate keyed to the stage at which the sublicence is granted, and — crucially — survival of sublicences on termination of the head licence.

Equity. The university takes a small equity stake in lieu of a portion of the up-front fee. Rafael negotiates anti-dilution language carefully; a technology transfer office's standard anti-dilution provision can complicate later financings.

Worked example two: the cross-licence

Halden Systems and Corvid Networks each hold portfolios in the same networking space and each has been accused by the other. Their general counsel agree to negotiate a cross-licence rather than litigate.

The negotiation is unlike the university licence in every respect.

Scope over rate. The parties agree quickly that no running royalty will be paid; the question is the balancing payment and the scope. Scope negotiation consumes the time: which patents, which products, which fields, and — most contentiously — whether the licence extends to future acquisitions.

Capture periods. Cross-licences typically cover patents with priority dates within a defined window. Halden wants a long capture period; Corvid, which is acquiring more aggressively, wants a short one. They settle on a five-year capture period with a mechanism for adding later-acquired portfolios by agreement.

Acquired entity provisions. If Corvid acquires a company with a large portfolio, does Halden get a licence to it? The standard answer is a limited one: the licence extends to the acquired entity's patents only as to products the acquired entity sold before the acquisition, and to Corvid's products only to the extent of pre-acquisition revenue. These provisions are intricate and worth drafting slowly.

Change of control. If Halden is acquired by a patent assertion entity, Corvid wants the licence to survive. It does, and both sides want it to.

Customer protection. Each party's customers are covered as to the licensed products. This is where exhaustion analysis enters, and both sides understand that a component sale under the cross-licence exhausts downstream claims.

Defensive suspension. If either party asserts patents outside the cross-licence against the other, the licence may be suspended. A useful stabilizer, and one that requires careful drafting so that a routine trademark dispute does not blow up a networking cross-licence.

Worked example three: the settlement licence

Bellweather IP has sued Ridgeline Manufacturing on two patents. Trial is nine weeks away. Both sides want out.

The pressures here are different again.

Speed. The agreement will be drafted in days. That is a reason to use a well-structured template, not a reason to skip the provisions that matter.

The comparable problem. Bellweather has thirty other targets. Whatever rate it accepts from Ridgeline becomes evidence in every subsequent negotiation and every subsequent trial. Bellweather will therefore resist a low headline rate and may prefer a lump sum with no rate stated — and will want the agreement to recite that it is a compromise of disputed claims, entered into to avoid litigation cost, and not evidence of a reasonable royalty.

The release and the covenant. Ridgeline needs a release for past conduct and a licence or covenant for the future. It also needs its customers and suppliers covered, and should ask explicitly. Bellweather should understand, per the exhaustion cases, exactly what a customer covenant does to its claims against the rest of the industry.

Scope of the release. Ridgeline should seek a release covering the full portfolio and all family members, not just the two asserted patents; Bellweather will resist. A common landing point is the asserted patents plus all patents claiming priority to the same applications.

No-challenge. Bellweather will want a covenant not to challenge validity. In a settlement of actual litigation where validity was contested, this is generally enforceable, and Ridgeline should price it rather than assume it is unenforceable.

Confidentiality, and its limits. Both sides want the terms confidential. Both should understand that the agreement will very likely be produced in Bellweather's next case under a protective order, and should draft with that reader in mind.

After signature: administering the licence

A patent licence is not a document you file. It is an obligation you run, usually for a decade or more, and the administration determines whether the negotiated terms produce the negotiated result.

Build the licence summary and circulate it. Finance needs to know what to report and when. Engineering needs to know the field limitation, in terms an engineer can apply. Sales needs to know the territory and channel constraints. Supply chain needs to know the have-made limitations. A licence that lives only in the legal department is a licence that will be breached by people who never saw it.

Set the calendar at signature. Report due dates. Payment due dates. Minimum measurement dates. Milestone deadlines and the internal work needed to hit them. Audit windows. Diligence deadlines. Patent expiration dates and the corresponding rate step-downs. Renewal and notice deadlines. Insurance renewals. Put a named owner on each.

Instrument the reporting. Royalty reports are generated from systems, and the systems have to be configured to produce the fields the licence requires. Configure them once, at the start, with the licence in hand. Retrofitting a royalty report from an ERP system three years in is a project; setting it up at the outset is a ticket.

Monitor the field boundary. If the grant is field-limited, somebody has to notice when the product roadmap approaches the boundary. Review the roadmap against the field definition annually and escalate before engineering has committed to a design.

Watch for change events. A licensor's assignment of the patents, a licensee's change of control, a co-owner's separate licence, an inter partes review, an acquisition on either side — each can change what the licence means. Set up a periodic check rather than relying on someone to remember.

Keep the file. The executed agreement with all amendments. The patent schedule as it stood at signature and as amended. Every report and payment. Every audit and its resolution. Correspondence on scope questions. Notices given and received. When the licence is diligenced in an acquisition — and it will be — this file is the difference between a clean disclosure schedule and a purchase price adjustment.

A note on rates, and why comparables mislead

Every negotiation eventually produces someone saying "the industry standard is X percent." There is no industry standard, and the comparables usually being cited are not comparable.

Published royalty rate surveys aggregate deals of wildly different character: exclusive and non-exclusive, single-patent and portfolio, early-stage and commercialized, arm's-length and settlement, with and without know-how, with and without minimums. The median of that set tells you almost nothing about the deal in front of you.

What actually determines a defensible rate:

  • The profit attributable to the patented feature, not to the product. A patent covering a component of a product should be priced against the component's contribution.
  • The next-best alternative. What would the licensee do without the licence, and what would it cost?
  • The remaining term. A patent with three years left is a different asset from one with twelve.
  • The strength of the position. Claim scope, validity risk, and the licensee's non-infringement arguments all discount the rate, and both sides know it.
  • Exclusivity. An exclusive licence in a defined field is worth a multiple of a non-exclusive one, and carries obligations to match.
  • What else comes with it. Know-how, materials, technical assistance, and improvements have independent value and, usefully, can support royalty obligations that survive patent expiration.

Where genuine comparables exist — the licensor's own prior licences in the same field, on similar terms, to similarly situated licensees — they are the best evidence available, and a licensee should ask for them. A licensor that has granted eleven licences at two percent and is asking for six will have to explain the difference, in the negotiation and later in court.

Four traps that recur

The licence that does not cover the product you are about to launch. A field defined around today's product, signed by a company whose roadmap will leave the field in two years. The fix is to negotiate the expansion mechanism at the outset — an option on adjacent fields at a stated rate is cheap when nobody wants it and expensive when everybody does.

The sublicence that dies with the head licence. A licensee grants a sublicence, the head licence terminates for reasons having nothing to do with the sublicensee, and a business built on the sublicence evaporates. The provision that prevents it is three sentences long and belongs in every sublicence: on termination of the head licence, the sublicence converts to a direct licence from the licensor on the same terms, provided the sublicensee is not in breach.

The improvements clause that transfers the company. A grant-back drafted broadly enough to capture everything the licensee develops in the field, exclusively, in perpetuity. The licensee spends five years and forty million dollars improving the technology and discovers it owns none of it. Define improvements narrowly, keep the grant-back non-exclusive, and make sure the definition cannot swallow independently developed work.

The royalty obligation that outlives the patents. A single rate running for a fixed twenty-year term with no reference to expiration. Under the Brulotte rule this is unenforceable as to the post-expiration period, which sounds like a licensee's windfall until you notice that the licensee has been paying it for four years and cannot recover what it paid. Schedule the step-downs.

What a good licence looks like when it is finished

You can tell a well-negotiated patent licence by a handful of characteristics.

The patent schedule lists numbers, with expiration dates. The grant clause enumerates the specific rights conveyed and states the field as a limitation on the grant rather than as a covenant. Have-made rights are present if the licensee uses contract manufacturers, and bounded if it does. Sublicensing is addressed explicitly, including survival. The royalty base is defined with an exhaustive deduction list. Rates step down by reference to stated dates. Reports specify their contents. The audit clause shifts cost at a stated variance. Improvements are defined narrowly, and grant-backs are non-exclusive. Enforcement rights and standing are addressed. Termination has notice and cure, and survival is specified. And there is a one-page plain-language summary attached, so that in six years someone can find out what the deal was without reading forty pages.

None of that is difficult. All of it takes attention at the moment when the commercial teams have shaken hands and everyone wants the lawyers to finish. The week spent then is the cheapest week in the life of the agreement.

One last thing: read the licence out loud to a non-lawyer

The final quality check costs twenty minutes. Take the executed draft to someone on the business side who was not in the negotiation, and walk them through it in plain English: here is what we can do, here is what we cannot, here is what we owe and when, here is what ends the deal.

Two things happen. Sometimes they say "that is not what we agreed," and you have caught a drafting error while it is still an error rather than a dispute. More often they ask a question nobody considered — what happens if we spin out the division, what happens if the supplier goes under, what if we want to sell in Brazil — and you find a gap that a year of careful drafting missed because everyone in the room shared the same assumptions.

A patent licence is a technical instrument, but the people who will live under it are not technicians of contract law. If it cannot be explained simply, it will not be followed accurately, and a licence that nobody follows accurately is a breach waiting for an audit.

Related documents