Royalty Audit and Licence Compliance Toolkit: Clauses, Scope, Findings, and Settlement

By Casey Scott McKay · 2025-10-30

Most licensors are underpaid, most know it, and most do nothing, because the audit right sits in a clause nobody has read since signature. This toolkit assembles the working material for licensors running audits, licensees receiving them, and practitioners drafting the clauses that make either possible. It covers the audit clause term by term and what each one costs when drafted badly, the desk review that finds most problems without any access at all, the records request that determines whether an audit can work, the coverage and deduction testing where the money actually is, and the settlement structure in which forward terms matter more than historic recovery. It closes with clause language, an authorities table, and the failures that recur on both sides.

IP and Technology > General IP | Toolkit | Published 30 October 2025 - Updated 4 July 2026 | Casey Scott McKay - marksy.us

Summary. Most licensors are underpaid and most do nothing about it. This toolkit covers the audit clause term by term, the desk review that finds problems without access, auditor engagement and the records request, coverage and deduction testing, the draft report and response cycle, the settlement structure in which forward terms outweigh historic recovery, and the licensee-side discipline that makes an audit survivable.

Keywords: royalty audit · audit clause · records definition · look-back period · cost shifting · acceptance clause · desk review · effective rate · auditor engagement · records request · coverage testing · deduction analysis · apportionment · draft report cycle · settlement forward terms · compliance programme

Start Here

Royalty auditing is a contractual right exercised through an accounting process and resolved as a legal negotiation, and it fails whenever one of those three disciplines is neglected.

The clause defines what is possible. Everything achievable is set out in a provision drafted years earlier, usually by somebody optimising a different term, and read by nobody since. The records definition in particular decides whether an audit can work at all.

Most underpayment is not fraud. It is drift: a product renamed, an entity acquired, a deduction category that crept in, a bundle apportioned by a methodology nobody documented, a currency convention applied inconsistently. Approaching an audit as an accusation guarantees a defensive response to a problem that is usually administrative.

The money is in coverage, the argument is in calculation. Whether a product, entity, territory, or channel is in the base at all is a bigger question than whether the arithmetic on the included ones is right — and it is answered faster.

And the forward terms are worth more than the recovery. A settlement that collects historic royalty and leaves the methodology untouched guarantees the same conversation in three years.

Four questions organise the work.

What does the clause permit?

Where is the exposure, and is it worth auditing?

What must the records request ask for?

And what does the settlement have to change?

See The Number on the Statement for the doctrinal treatment, Conducting or Defending a Royalty Audit for the sequence, and the Royalty Audit Checklist for the working list.

Part one: the audit clause, term by term

Trigger and notice. Who may exercise, what notice, in what form, and how often. Check whether an affiliate or a successor may audit, because assignment histories break this regularly. Check whether the right survives termination and for how long, since a licensor auditing a terminated licence with no survival clause has nothing.

Look-back. How many years the audit may reach. Compare it against the contractual and statutory limitation periods and against the audit frequency, because a three-year look-back with a three-year cycle means nothing is ever forfeited and a two-year look-back with a four-year cycle means periods expire unaudited.

Records definition. The single most important term. "Books and records relating to the calculation of royalties" confines the auditor to the licensee's own workings and is close to useless. A workable definition reaches sales ledgers, product master data, customer and distributor agreements, system extracts, and the documentation supporting each deduction category.

Auditor. Who may conduct it, whether the licensee approves, whether approval may be unreasonably withheld, and whether contingency-fee auditors are prohibited, which many clauses do and which constrains the licensor's cost structure.

Access. Premises, hours, duration, and whether remote or system access is permitted. A clause contemplating a visit to a filing room does not contemplate an extract from an enterprise system.

Cost and shifting. Who pays as default, and the discrepancy threshold at which the licensee pays. Five per cent is broadly market; higher thresholds are a gift. Whether the threshold applies to the audited period as a whole or to any single period matters.

Interest. The rate and the date from which it runs.

Acceptance and deemed approval. A provision making statements final after a stated period is the most valuable term a licensee can obtain and the most damaging a licensor can concede. Its removal at renewal is the highest-value single change available.

Dispute resolution. Audit findings frequently go to expert determination rather than to the licence's general dispute clause, and the distinction should be deliberate.

Reporting format. Not strictly an audit term and the one that prevents most disputes: gross, deductions by category, net, rate, royalty, by product and by territory.

Notification obligations for new products and new entities, which address the two largest error categories prospectively.

And an annual certification by an officer, which costs the licensee nothing and creates the internal accountability that prevents drift.

Part two: the desk review

Between doing nothing and commissioning an audit sits the highest-return activity in the discipline, and most licensors do not run it.

What it is. A structured review of the statements and public information, conducted in a day or two, with no access to the licensee's records.

Effective rate. Royalty received divided by reported sales, compared to the contractual rate, tracked across periods. A movement of half a percentage point without a contractual change is a question.

Deduction percentage over time. A rising proportion is the clearest single indicator of methodology drift.

Product coverage. The licensee's public catalogue against the products appearing in the statements.

Entity coverage. The group's corporate structure, from filings and announcements, against the entities reporting. Acquisitions are the recurring gap.

Territory and channel coverage against the licensee's own marketing: marketplaces, direct-to-consumer, subscription, and bundled distribution.

Restatements, which indicate a calculation nobody is confident in.

Arithmetic. Recompute the face of each statement. Errors in addition are more common than anybody expects.

What it produces. Comfort, a specific question worth asking, or a targeted audit — and a targeted audit is substantially cheaper than a general one because the auditor is looking for something.

The procedural benefit. A licensor that raises a specific query in writing and receives an explanation has preserved arguments that silence would waive and has begun a record that defeats a later acceptance or course-of-dealing defence.

And it should be annual across the portfolio. A desk review over fifteen licences is a week's work a year, costs less than one audit, and tells the licensor where the audits should go.

Part three: engagement and the records request

Select the auditor for sector experience and systems capability, not for brand. An auditor who has audited licences in the sector knows the deduction categories; one who has not will learn on the licensor's money.

Check conflicts against the licensee's group before approaching, and seek the licensee's approval where the clause requires it.

Ask about sampling before engaging. A firm that samples where full population testing is possible will produce a weaker report, because extrapolation is the licensee's most reliable methodological defence.

Engage on terms that make the report usable: findings expressed by reference to the specific licence provision; interpretive assumptions listed separately from arithmetic findings; modular quantification so a conceded category can be removed; documented data lineage; and a draft to the licensor before anything goes to the licensee.

Consider privilege. Engaging the auditor through counsel affects the position in some jurisdictions, and interpretive questions should be routed through counsel rather than resolved by the auditor.

Then design the records request, which decides whether the audit works.

System and master data. Chart of accounts, product master file with every field used to determine royalty-bearing status, the mapping logic and its change log, customer and channel master data, the entity list, and the royalty calculation documentation including manual steps.

Transaction data. Invoice-line granularity, machine-readable, with date, entity, customer, country of sale, product code, description, quantity, gross price, each deduction type separately, net amount, currency, and rate applied. Credit notes, returns, and rebates separately, linked to the original invoices.

Reconciliation. Reported net sales to the ledger, and the ledger to the statutory accounts for at least one year.

Deduction support. The policy for each category, transaction-level support for the largest, the freight and insurance basis and allocation method, rebate agreements, the bad debt policy, and the true-up mechanism for any estimated deduction.

Contracts and coverage. Distributor and reseller agreements, sublicences and sublicensee reporting, bundle pricing documentation and apportionment methodology, marketing materials and price lists, and details of any acquisition during the period.

Anticipate the objections and answer them in writing with the contractual basis, recording every request, refusal, and partial production, because the pattern matters if the matter escalates.

Part four: testing

Reconcile before testing anything. An auditor who tests transactions without tying reported royalty to the ledger is testing an unverified population, and the findings will not survive.

Test coverage before accuracy. Product, entity, territory, and channel.

Product coverage. Every product excluded from the base, with the licensee's stated reason. Test successor and variant products specifically, since a product renamed after a redesign is the most common coverage error. Test service and subscription offerings that embed the licensed subject matter, spare parts, consumables, and accessories.

Entity coverage. Every group entity selling licensed product that does not report. Test intra-group sales and identify where the royalty is calculated — on the transfer price or on the ultimate third-party sale. Test entities acquired during the period.

Territory and channel. Marketplace sales reported net of platform commission where the licence says gross; direct-to-consumer sales in a separate system; bundled and promotional distribution; free-of-charge units where the licence provides a deemed price.

Then test calculation. Rate application including tiers, stacking and offset provisions, most-favoured-licensee clauses, and minimums.

Deductions. Every category applied, matched against the licence's permitted list. Flag every category applied that does not appear in the licence, which is the single most common finding. Test quantum against support, estimated versus actual with true-up, freight allocation, cash discounts against actual payment terms, volume rebates against trade agreements, returns for double counting, and bad debt for subsequent recovery. Test whether any deduction is applied to the royalty rather than to the base, which changes the arithmetic materially.

Apportionment. Every bundle, suite, subscription, or platform offering containing licensed and unlicensed elements, with the methodology, its consistency, its adoption date, and the gap between it and a defensible alternative — so the settlement discussion has a range.

Currency and timing. Exchange rate source and conversion date, revenue recognition point against the licence's definition of a sale, period cut-off at each year end, and withholding tax against the licence and the treaty position.

Methodology. Prefer full population testing; stratify where sampling is necessary; quantify each finding independently; classify each as clear error, interpretive difference, or coverage question in the working papers; document the basis by reference to the provision and the data; and assess the licensee's own controls, since a documented mapping with an annual review predicts fewer errors and tells the auditor where to concentrate.

Part five: the report cycle and the settlement

Read the draft before the licensee does. Draft reports overstate, because auditors resolve ambiguity in favour of the party paying them. Check that every finding is provision-referenced, that interpretive assumptions are visible, that quantification is modular, and that the tone does not characterise interpretive differences as underreporting.

Form a settlement range before the draft goes out, so the negotiation is not conducted reactively.

Give a realistic response period. A licensee given ten days produces a rejection rather than an explanation.

Anticipate the defences: limitation on the earliest periods; any acceptance or deemed-approval clause; course of dealing built from years of unqueried statements; methodology attacks on sampling and extrapolation; contra proferentem on any ambiguous definition the licensor drafted; and a scope objection that the auditor exceeded the audit right.

Expect a counterclaim in adversarial matters, including a validity challenge, which a licensee may bring without breaching following Lear, Inc. v. Adkins and MedImmune, Inc. v. Genentech, Inc..

Concede clearly wrong findings promptly, in writing. It costs nothing and buys credibility on the rest.

Settle category by category rather than as a lump, so agreed findings are not discounted by disputed ones.

And make the forward terms the priority: the methodology change in writing for each interpretive finding; the product mapping and entity list going forward; the permitted deduction categories, with the unauthorised ones deleted; the bundle apportionment methodology with a worked example annexed; the improved statement format; the new product and new entity notification; the annual certification; and a follow-up review to confirm implementation.

Define the release narrowly by period and subject matter, excluding matters outside the audit scope, excluding fraud, and preserving the go-forward methodology.

And if it will not settle, model the arithmetic honestly, since the cost of reaching trial routinely exceeds the disputed sum, and confirm the dispute route the clause requires.

Part six: the licensee's programme

A licensee that treats audits as an event will be audited badly. One that treats compliance as a process will be audited rarely and cheaply.

Assign ownership. A named person accountable for the royalty calculation, with a documented annual review.

Document the mapping. Which products are royalty-bearing, on what basis, with a change log. This single document answers most of an audit.

Re-read the definitions annually, particularly Licensed Product, Net Sales, and the permitted deduction list, and test the current practice against them rather than against last year's practice.

Test the deduction categories against the licence and stop applying any that does not appear in it, because that is the first finding in every audit.

Document the apportionment methodology for any bundle, with a worked example, and apply it consistently.

Capture new products and new entities through a process rather than by hope, since acquisitions and product renames are the two largest error categories.

Reconcile before reporting, tying the statement to the ledger every period rather than at audit time.

Report in a readable format — gross, deductions by category, net, rate, royalty — because a transparent statement invites fewer questions than an opaque one.

Assemble deduction support as you go. Documentation gathered under audit pressure produces gaps that read as concealment.

Respond to audits professionally: a single named contact, a log of every request and response, written objections with the contractual basis where a request exceeds the records definition, and briefed personnel who answer accurately, briefly, and within their knowledge.

Trade historic recovery for forward clarity in settlement, which is usually a good trade for a licensee that wants certainty.

And fix the underlying process, because the next audit is coming and the same errors will otherwise be found again.

Clause bank

Records definition. "Records" means all books, records, and data relating to the manufacture, distribution, sale, licensing, or other disposition of Licensed Products, including: sales ledgers and general ledger accounts; transaction-level sales data at invoice-line granularity in machine-readable form; product master data and the mapping identifying royalty-bearing products, with its change log; customer, distributor, and channel master data; distributor, reseller, and sublicence agreements; documentation supporting each deduction applied; bundle and platform pricing documentation and the apportionment methodology; the workings underlying each Royalty Statement; and the reconciliation between reported Net Sales and Licensee's statutory accounts. Records include those held by any Affiliate and by any third party on Licensee's behalf.

Audit right. Licensor may, on [30] days' notice, not more than once in any twelve-month period, appoint an independent accountant to audit the Records for any period ending not more than [X] years before the notice. Licensee shall provide access to the Records, including remote access to systems and the production of data extracts in the format specified, and shall procure equivalent access from each Affiliate. This right survives termination for [X] years. If the audit discloses an underpayment exceeding [5] per cent for any audited period, Licensee shall pay the reasonable cost of the audit in addition to the underpayment and interest at [rate] from the date each amount fell due.

Reporting format. Each Royalty Statement shall set out, by Licensed Product, by territory, and by Affiliate: gross invoiced amounts; each deduction applied, itemised by category; Net Sales; the applicable rate; and the royalty payable. Where any amount is converted from another currency, the statement shall state the source and date of the exchange rate applied. Where any Licensed Product is sold as part of a bundle, the statement shall state the amount attributed to the Licensed Product and the methodology applied.

Notification and certification. Licensee shall notify Licensor within [30] days of: introducing any product that is or may be a Licensed Product; any entity commencing sale of Licensed Products; any change to the methodology by which Net Sales or any deduction is calculated; and any change to the apportionment methodology for any bundle. Licensee shall deliver annually, within [60] days of its financial year end, a certificate signed by an officer confirming that royalties have been calculated in accordance with this Agreement and that the mapping of royalty-bearing products is complete and current.

No acceptance. No Royalty Statement shall be treated as accepted, approved, or final by reason of the passage of time, the absence of query, the receipt of payment, or any course of dealing. Licensor's failure to query any statement shall not waive any right in respect of it or any other statement.

Settlement forward terms. With effect from [date], the parties agree that: the products listed in Schedule [A] are Licensed Products; the entities listed in Schedule [B] shall report; the deductions permitted are those listed in Schedule [C] and no others; bundles shall be apportioned in accordance with the methodology and worked example at Schedule [D]; statements shall be delivered in the format at Schedule [E]; and Licensee shall provide the notifications and certification at clause [X]. Licensor shall conduct a follow-up review of the first two statements delivered after this date, at its own cost, limited to confirming implementation of these terms.

Worked scenarios

The deduction that crept in. A licensee begins deducting a distribution allowance in year four. It is not in the permitted deduction list. Nobody decided to breach the licence; a finance manager applied the company's standard net sales definition when a system was upgraded. By year seven the allowance is four per cent of gross and the cumulative shortfall is substantial. The audit finds it in an afternoon. The settlement recovers part of the history and — more importantly — deletes the category going forward, which is worth more than the recovery over the remaining term.

The acquisition nobody reported. A licensee acquires a competitor whose product line falls within the Licensed Product definition. The acquired entity continues reporting to its own systems and never enters the royalty calculation. Three years of sales are unreported. The licensor's desk review would have found it from a corporate announcement in month one, and the notification obligation in the clause bank above would have prevented it entirely.

The bundle. A software licensee launches a suite containing the licensed component and four others, and apportions twelve per cent of the bundle price to the licensed component on a methodology nobody documented. The licence is silent on apportionment. The auditor's alternative methodology, based on standalone selling prices, produces thirty-one per cent. Neither number is obviously right, and the settlement lands in between with a documented methodology and a worked example annexed — which is the outcome that actually matters, because the bundle will be sold for another decade.

The acceptance clause. A licensor discovers a significant underreporting in years two to five. The licence provides that statements become final ninety days after delivery absent written objection. No objection was ever made. Four years of exposure evaporate. The clause was conceded during a negotiation about something else, by somebody who did not know what it did.

Failures that recur

Licensor side. Auditing too late, after limitation has taken the earliest years. Auditing everyone equally rather than by exposure. Treating the draft report as the answer. Ignoring the forward terms. Failing to feed findings back into the template. Confusing silence with acceptance and thereby building the licensee's course-of-dealing defence. Conceding an acceptance clause. Accepting a narrow records definition. Never running a desk review.

Licensee side. No named owner for the royalty calculation. An undocumented product mapping. Deduction categories inherited from a general net sales definition rather than from the licence. An apportionment methodology nobody wrote down. Acquisitions that never enter the calculation. Documentation assembled under audit pressure. Answering an audit reactively rather than having run the numbers first.

Both sides. Treating the audit as an accusation rather than as a reconciliation of a calculation that has drifted, which is what it almost always is.

Part seven: variations by rights type

The framework is general; the exposure concentrates differently by subject matter.

Patent licences. Coverage is the dominant question, and it turns on claim scope: whether a product falls within the Licensed Product definition depends on whether it practises the claims, which is a technical analysis the auditor cannot perform alone. Bring a technical adviser. Post-expiry royalty questions arise under Brulotte v. Thys Co. where the portfolio contains expiring families, and portfolio licences need a mechanism for what happens as patents lapse.

Trademark licences. Coverage questions concern which goods bear the mark and which channels are captured; quality control failures surface alongside underreporting; and the accounting remedy under 15 U.S.C. § 1117 is available where the relationship breaks down. See Naked Licensing.

Copyright and content licences. Usage-based reporting, platform-mediated distribution, and territory restrictions dominate, with the recurring problem that the licensee's own reporting depends on data supplied by a platform it does not control.

Software licences. Seat, instance, and deployment counting rather than revenue; virtualisation and container deployment defeat licence models written for physical servers; and the audit is a deployment audit rather than a financial one.

Know-how and hybrid licences. Apportionment between patent and know-how components matters for post-expiry royalties and for tax, and the licence should state it. See Aronson v. Quick Point for the underlying position.

Data licences. Derived data definitions, redistribution, and internal versus external use, with the same audit and back-payment exposure described in the Data Licensing Checklist.

Character and franchise licences. Merchandise categories, territory, and the treatment of promotional and free-of-charge goods, with the additional feature that licensees frequently sublicense. See the Character and Franchise Rights Checklist.

And university and research licences, where diligence obligations, milestone payments, and sublicensing income sharing sit alongside the running royalty, and where the licensor's audit right is frequently the only mechanism for verifying any of it. See the Technology Transfer Checklist.

Part eight: the portfolio programme

A licensor with more than a handful of licences should run a programme rather than a series of events.

Rank annually. Every live licence with its base, its contractual rate, its effective rate over three years, and the flags from the desk review.

Audit on a rotation and tell licensees the rotation exists. An audit that is expected is a routine matter; one that arrives unannounced after eight years is an accusation.

Budget it. Two audits a year from a portfolio of fifteen, plus an annual desk review across all of them, is a predictable cost with a predictable return.

Feed every finding into the template. A deduction error at one licensee is usually present at three, and the template change prevents it at the next twenty.

Diary every renewal, because renewal is the only moment these terms improve without being paid for. Lead with the records definition, framed as clarification; extend the look-back; move the threshold to five per cent; add the reporting format, the notification obligations, and the certification; and remove any acceptance clause.

Track implementation of settlement forward terms against the first two statements after each settlement, and record any drift immediately.

Report to the board annually: royalties received, effective rates, audits conducted, recoveries obtained, forward terms secured, and template changes made.

And measure the right thing. The programme's return is not the recovery. It is the recovery plus the improvement in every subsequent year's reporting across every licence the template touched, which is the number that justifies the function.

Part nine: pricing the engagement

Clients ask what an audit costs, and the components behave differently.

The clause review and triage is a small fixed cost and should be done across the portfolio at once rather than per engagement.

The auditor's fee scales with the licensee's complexity rather than with the licensor's revenue. A single-entity licensee on one system costs a fraction of a multinational on six.

Legal cost during fieldwork is low, since the work is accountancy. It rises sharply at the interpretive stage, where the definitions are argued.

The response and settlement phase is unpredictable, depending entirely on how much sits in the interpretive category.

Cost shifting changes the arithmetic where the threshold is met, which is why the threshold's level deserves more negotiating attention than most licensors give it.

And the return is not only the recovery. The forward terms, the template corrections, and the improvement in every other licensee's reporting are the compounding part and belong in the business case.

The honest framing for a first engagement is that the first audit may or may not recover its cost and the programme almost certainly will — because subsequent audits are cheaper, better targeted, and conducted against licensees who now know the cycle exists.

Part ten: when it becomes a dispute

Most audits settle; the minority that do not follow a recognisable path.

The claim is contractual, for underpayment, with interest and, where the clause provides, costs.

Discovery reaches into the licensee's financial systems, which is expensive, intrusive, and frequently the reason a licensee prefers to settle. See the IP Discovery Checklist.

Expert accounting evidence is required on both sides for anything beyond arithmetic, and the experts will disagree about apportionment and deduction boundaries rather than about the numbers.

Contract interpretation is the core, with contra proferentem available against the drafter of any ambiguous term.

Limitation is litigated early, since it may dispose of most of the claim without reaching the merits.

Counterclaims appear, including validity challenges and allegations of licensor breach.

Termination is rarely in the licensor's interest, since a terminated licence produces no royalties.

And the arithmetic of proceeding is usually unfavourable, because the cost of reaching trial routinely exceeds the disputed sum. That should inform the settlement discussion openly rather than being discovered after the pleadings.

One paragraph to remember

Read the clause before spending anything, because the records definition decides whether an audit can work. Run a desk review across the whole portfolio annually — it costs a week and it finds most of what an audit would. Engage an auditor who will express findings by licence provision and quantify them modularly. Reconcile before testing, test coverage before accuracy, and expect the money to be in products, entities, territories, and channels rather than in arithmetic. Concede the wrong findings promptly. And treat the settlement's forward terms as the deliverable, because the historic recovery is a payment and the corrected methodology is an annuity.

The auditor's report as a legal document

A royalty audit report is written by an accountant and read by lawyers, and the gap between those audiences causes more difficulty than any other feature of the process.

Ask for findings expressed by licence provision. "Underreported net sales of $1.4m" is unusable. "Products X, Y and Z fall within the definition of Licensed Product at clause 1.12 because they incorporate the claimed method, and were excluded from the base, giving rise to unpaid royalty of $1.4m" can be argued and, if necessary, pleaded.

Ask for interpretive assumptions to be listed separately. Every audit rests on a reading of the definitions; where the auditor has adopted one, it should be visible, because the licensee will contest it and the licensor needs to know which findings depend on it.

Ask for modular quantification. If the freight deduction finding is conceded, the licensor must be able to remove it without recalculating everything.

Ask that "should have been" language be avoided where the licence is genuinely ambiguous, because it converts a negotiable interpretive question into an accusation.

Ask for data lineage — which extract, which fields, which reconciliation — since a finding whose source cannot be retraced eighteen months later will be abandoned.

And read the draft before it goes out. Counsel reviewing the draft for tone and characterisation is the cheapest risk reduction available in the process, and it is skipped constantly.

Documents that must exist

For each item: does it exist, who owns it, and can it be produced on request?

The clause summary: one page per licence, recording trigger, notice, look-back, records definition, auditor conditions, access, cost shifting, interest, acceptance, and dispute route.
The portfolio ranking, updated annually, with effective rates over three years.
The desk review record per licence, with the questions raised and the answers received.
Written queries sent to licensees and their responses, which defeat course-of-dealing arguments.
The auditor engagement letter with the report-quality requirements.
The records request as served, and the log of every request, refusal, and partial production.
The reconciliation working papers.
The coverage testing record: products, entities, territories, channels, with exceptions and the provision each engages.
The deduction analysis with the permitted list and the applied list side by side.
The apportionment analysis with the licensee's methodology and the alternative.
The draft report, the licensor's review comments, the licensee's response, and the final report.
The settlement agreement with the forward terms schedules.
The follow-up review of the first two statements after settlement.
The template change record, showing which findings changed the standard licence and why.
The renewal diary.

On the licensee side: the product mapping with its change log; the deduction policy tested against the licence; the apportionment methodology with a worked example; the reconciliation performed each period; the named owner and the annual review record; and the audit response log.

A closing note on the relationship

The reluctance to audit is almost always framed as concern for the relationship, and it deserves a direct answer.

A badly conducted audit does damage relationships: an aggressive auditor, an inflated draft report, and a licensor treating interpretive findings as accusations. But the alternative is not a preserved relationship. It is a licensor who does not know what it is owed, a licensee whose calculation drifts further from the contract every year, and a shortfall that eventually grows large enough to force a confrontation on far worse terms.

Audits run routinely, on a known cycle, by a professional auditor, by a licensor that distinguishes clear error from interpretive disagreement, are treated by sophisticated licensees as ordinary commercial hygiene. They are frequently welcomed, because the licensee's own finance team has been uncertain about the calculation for years and has had no mechanism for resolving it.

The damage comes from the surprise. Which means the answer is to audit more often rather than less, and to say in advance that you will.

The one-page test

The quickest diagnostic on a licensing programme takes an hour. Pick the licence generating the most royalty and ask five questions.

What is the records definition, in its exact words, and would it reach a transaction-level extract from the licensee's system? What was the effective rate in each of the last three years, and has it moved? Which entities in the licensee's group report, and does that list match the group structure in its public filings? Which deduction categories appear on the statements, and do all of them appear in the licence? And when was this licensee last audited, and what does the clause say about how far back an audit can now reach?

A licensor that answers all five in an hour is running a programme. One that answers two is the ordinary case. One that cannot answer the first has an audit right it has never read, protecting a revenue stream it cannot verify — which describes most licensors, and which is why this practice exists.

Key Authorities at a Glance

Licence and royalty framework. 35 U.S.C. § 261 on patent assignment and licensing; 17 U.S.C. § 204 on copyright transfers; 15 U.S.C. § 1055 on related company use and 15 U.S.C. § 1127 on abandonment for trademark licences.

Royalty limits and structure. Brulotte v. Thys Co., reaffirmed in Kimble v. Marvel Entertainment, LLC, on post-expiry patent royalties; Aronson v. Quick Point Pencil Co. on hybrid know-how royalties; 35 U.S.C. § 154 on term.

Exhaustion. Impression Products, Inc. v. Lexmark International, Inc.; Kirtsaeng v. John Wiley & Sons, Inc.; Quanta Computer, Inc. v. LG Electronics, Inc..

Licensee challenges. Lear, Inc. v. Adkins; MedImmune, Inc. v. Genentech, Inc.; 28 U.S.C. § 2201.

Damages and apportionment, where an audit becomes a claim. 35 U.S.C. § 284; Georgia-Pacific Corp. v. United States Plywood Corp.; LaserDynamics, Inc. v. Quanta Computer, Inc.; Uniloc USA, Inc. v. Microsoft Corp.; 15 U.S.C. § 1117 with Romag Fasteners, Inc. v. Fossil, Inc. for trademark licences.

Procedure and evidence. Fed. R. Civ. P. 26 and Fed. R. Civ. P. 34 on discovery reaching financial systems; Fed. R. Evid. 702 with Daubert v. Merrell Dow Pharmaceuticals, Inc. on expert accounting evidence; Fed. R. Evid. 1006 on summaries of voluminous records.

Tax. 26 U.S.C. § 482 and 26 C.F.R. § 1.482-4 on transfer pricing, which interacts with intra-group base and withholding findings.

| Authority | Governs | Where it bites | | --- | --- | --- | | 35 U.S.C. § 261 | Licensing | Who may exercise the audit right | | Brulotte; Kimble | Post-expiry royalties | Periods after expiry are not recoverable | | Aronson | Hybrid royalties | Know-how components survive | | Lear; MedImmune | Licensee challenge | The counterclaim risk | | Impression Products | Exhaustion | Whether downstream sales bear royalty | | Kirtsaeng | International exhaustion | Grey-market units in the base | | 35 U.S.C. § 284 | Damages | The fallback if the licence fails | | LaserDynamics | Apportionment | Bundle allocation arguments | | 15 U.S.C. § 1117 | Trademark recovery | Accounting on a mark licence | | Fed. R. Evid. 1006 | Summaries | How findings reach a trier of fact | | 26 U.S.C. § 482 | Transfer pricing | Intra-group base and withholding | | Acceptance clauses | Finality | The single most damaging concession |