Joint Ventures and Strategic Alliances: Governance, Deadlock, Contributions, and Exit

Document type: Article Practice area: Corporate — Joint Ventures and Alliances Jurisdiction: United States (Delaware entity law and federal antitrust) Last reviewed: 5 September 2026

The structural problem

An acquisition ends. A supply contract has a term. A joint venture is supposed to last, and it puts two organizations — each with its own board, its own shareholders, its own strategic plan, and its own quarterly earnings pressure — in permanent partnership over a business that neither controls alone.

That is a genuinely unstable arrangement, and the failure rate reflects it. Ventures rarely fail because the business idea was wrong. They fail because:

The parties' strategies diverge. One partner wants to reinvest; the other wants distributions. One wants to expand into a market where the other already competes. One is acquired by a company that hates the venture.
Contributions turn out to be unequal in ways nobody priced. The partner who contributed cash finds that the partner who contributed "technology and market access" has stopped contributing either.
The governance deadlocks and there is no exit that anyone will actually use. A buy-sell provision that requires $400 million of cash is not an exit; it is a decoration.
The venture becomes more valuable to one partner than the other, and that partner starts behaving accordingly.

The drafting response is not to prevent divergence — you cannot — but to make its consequences predictable. A well-drafted venture agreement is a set of pre-negotiated answers to questions the parties will otherwise fight about when they are angry and their interests are opposed.

Choosing the vehicle

Contractual alliance versus entity

Not every collaboration needs an entity. A contractual alliance — co-marketing, co-development, distribution, a joint bid — allocates rights and obligations without creating a separate business. It is faster, cheaper, easier to unwind, and avoids the tax and consolidation questions an entity raises.

Use an entity when:

The collaboration needs to own assets — plant, licenses, intellectual property, a customer contract in its own name;
It needs to incur third-party liabilities and the parties want them ring-fenced;
It requires outside capital or debt in its own name;
Employees will work for it;
The parties want a defined equity stake that can be valued, transferred, or sold.

Use a contract when the collaboration is bounded in scope and time, when the parties want to keep their operations separate, or when antitrust or regulatory considerations counsel against a formal combination.

The most common structuring error is creating an entity for a relationship that is really a contract. An entity brings governance, capital accounts, tax filings, fiduciary questions, and an exit problem. If the parties are simply going to sell each other's products, write a distribution agreement.

Which entity

For a US venture, the practical choice is almost always a Delaware limited liability company, for three reasons.

Flexibility. The Delaware LLC Act gives the parties near-total freedom to design governance. Voting can be by class, by matter, or by veto. Managers can be appointed rather than elected. Economics can be separated entirely from control.

Pass-through taxation. A two-member LLC is a partnership for federal tax purposes by default, so income is taxed once and losses flow through to partners who may be able to use them. Corporate ventures generate a second layer of tax on distributions.

Fiduciary duty modification. Delaware permits an LLC agreement to expand, restrict, or eliminate fiduciary duties, provided the implied contractual covenant of good faith and fair dealing is not eliminated. This is the single most important structural feature for a joint venture, and it deserves its own section below.

Corporations remain appropriate where the venture will seek venture capital or go public, where a foreign partner's tax position favors a corporation, or where the parties want the predictability of well-developed corporate law rather than a bespoke contract.

Contributions: the valuation problem nobody solves cleanly

Every venture begins with an argument about what each side is putting in.

Cash is easy. Everything else is not.

Contributed assets — equipment, facilities, inventory — need valuation, and the valuation drives both ownership percentages and tax basis. Expect the contributing party's number to be optimistic.
Contributed intellectual property is the hardest. Is it assigned or licensed? Exclusively or not? For the venture's field only, or broadly? What happens to improvements? What happens on termination? A venture built on a license that terminates when the venture does has a fundamentally different value than one that owns the technology.
Contributed people are usually seconded rather than transferred, which means they keep their old employer's incentives, benefits, and loyalties. Secondment agreements should address who directs the work, who owns the inventions, and what happens at the end.
Contributed market access, relationships, and "commitment" are the contributions that most often fail to materialize. If a partner's contribution is that it will sell the venture's product through its sales force, that obligation belongs in the agreement with a metric, not in the recitals as a statement of intent.

The drafting response: contribution and commitment covenants

Convert soft contributions into hard obligations:

Minimum purchase or distribution commitments, with remedies.
Dedicated headcount, specified by role and full-time-equivalent.
Defined service levels for shared services provided by a parent.
Explicit IP grants, with field, territory, exclusivity, sublicensing, improvements, and survival all addressed.
Consequences for failure short of blowing up the venture: dilution, a reduction in board seats, loss of a veto, or a put right for the performing party.

Dilution as the default remedy is worth emphasizing. Damages for a failed contribution are hard to prove and destructive to the relationship. A formula that adjusts ownership when a party fails to fund or fails to perform is self-executing and proportionate.

Capital calls and the funding trap

Most ventures need money again. The agreement must answer:

Who can call capital, and on what vote? If a supermajority is required and one party can block, the venture can starve.
What happens if a party does not fund? Options include dilution on a formula (punitive or proportionate), a loan from the funding party at a stated rate convertible at a discount, loss of governance rights, or a forced sale.
Is there a cap? Partners with different balance sheets have different tolerance for open-ended commitments. A parent guarantee may be required from the smaller party.

The classic squeeze is a venture in which the deep-pocketed partner calls capital repeatedly, knowing the other cannot fund, and dilutes it to insignificance. The classic defense is a cap on mandatory capital, a punitive-dilution ceiling, or a right to convert to a non-diluting preferred position.

Governance: control without a controller

Board composition and the deadlock it creates

A 50/50 venture with an even board deadlocks by design. That is sometimes intentional — neither party will invest without a veto — but it means the deadlock-breaking mechanism is the most important provision in the document.

Common structures:

Structure	How it works	Risk
Even board, all decisions by majority	Deadlock on everything	Paralysis without a tiebreak
Even board, independent chair with casting vote	Neutral breaks ties	Finding a genuinely neutral chair; the chair becomes the real decision-maker
Uneven board reflecting ownership	Majority governs	Minority needs reserved matters
Rotating control	Control alternates by period	Discontinuity; gaming near transitions
Majority board, broad reserved matters for the minority	Ordinary course runs; significant matters need consent	The reserved-matter list is where the negotiation actually happens

Reserved matters

The reserved-matter (or "supermajority," or "consent") list defines what the minority controls. A well-constructed list is short and consequential. A badly constructed one is long and operational, which produces deadlock over routine decisions.

Belongs on the list:

Amending the LLC agreement or the certificate;
Issuing new equity or admitting a member;
Merger, sale of the venture, or sale of substantially all assets;
Dissolution;
Incurring indebtedness above a threshold, or granting liens;
Approving the annual budget and business plan (with a fallback if not approved);
Material changes in the scope or line of business;
Transactions with a member or its affiliates above a threshold;
Distributions policy;
Appointment and removal of the chief executive;
Commencing or settling litigation above a threshold;
Entering into or terminating material contracts above a threshold.

Does not belong: hiring below the executive level, ordinary purchasing, marketing decisions, and anything the venture must do dozens of times a year.

The budget provision is the sleeper. If the annual budget requires unanimous consent and there is no fallback, a partner can shut the venture down by refusing to approve. The fix is a deemed-budget provision: if no budget is approved by a date, the prior year's budget continues with an inflation adjustment and permitted variances. This single clause prevents more deadlocks than any dispute-resolution mechanism.

Fiduciary duties: modify them deliberately

Delaware LLC law permits fiduciary duties to be expanded, restricted, or eliminated. Ventures almost always modify them, because the default rules make ordinary partner behavior a breach.

Consider a venture between two industrial companies. Each appoints managers who are its own employees. Under default fiduciary principles, those managers owe undivided loyalty to the venture, which means the manager who reports to their employer about the venture's plans — the entire reason the employer appointed them — is breaching a duty.

Standard modifications:

Managers appointed by a member may act in the interests of that member, and owe no duty to the venture or the other member in doing so, except as expressly stated.
Members may pursue other business opportunities, including competing ones, and the corporate opportunity doctrine is expressly waived (or narrowed to a defined field).
Information sharing with the appointing member is permitted, subject to confidentiality obligations flowing to that member.
Affiliate transactions approved by a stated process are deemed fair.

Two limits. First, the implied contractual covenant of good faith and fair dealing cannot be eliminated, and Delaware courts use it to police conduct that defeats the parties' reasonable expectations without violating any express term. Gotham Partners, L.P. v. Hallwood Realty Partners, L.P., 817 A.2d 160 (Del. 2002) illustrates the analysis in the alternative-entity context: where the agreement supplies a standard, that standard governs, and the court's task is contract interpretation.

Second, modification must be express. Auriga Capital Corp. v. Gatz Properties, LLC, 40 A.3d 839 (Del. Ch. 2012), affirmed on other grounds in Gatz Properties, LLC v. Auriga Capital Corp., 59 A.3d 1206 (Del. 2012), involved a manager who ran an auction designed to deliver the entity to himself cheaply. The agreement did not clearly eliminate duties, and the manager was held liable. Silence is not waiver. If the parties want duties gone, the agreement must say so unmistakably.

Contrast the corporate context, where fiduciary duties cannot be waived and the exclusive-benefit test of Sinclair Oil Corp. v. Levien, 280 A.2d 717 (Del. 1971) governs parent-subsidiary dealings. That difference alone drives most ventures to the LLC form.

The antitrust overlay

Two competitors forming a venture are doing something the Sherman Act watches closely. The analysis has three parts.

Is the venture a single entity or an agreement between competitors?

Copperweld Corp. v. Independence Tube Corp., 467 U.S. 752 (1984) held that a parent and its wholly owned subsidiary are a single economic actor incapable of conspiring under 15 U.S.C. § 1. Joint ventures owned by competitors are different.

American Needle, Inc. v. National Football League, 560 U.S. 183 (2010) rejected a categorical single-entity defense for a licensing venture owned by thirty-two competing teams. The test is functional: does the arrangement join separate economic actors pursuing separate economic interests, such that it deprives the marketplace of independent centers of decision-making? A venture whose owners remain actual or potential competitors in the relevant market is subject to § 1 scrutiny.

Is the venture itself lawful, and are its restraints ancillary?

Legitimate integrations are analyzed under the rule of reason, and restraints reasonably necessary to the integration are analyzed with it rather than condemned separately.

Broadcast Music, Inc. v. Columbia Broadcasting System, Inc., 441 U.S. 1 (1979) is the foundational case: a blanket license set by a venture of competing composers was literally price fixing, but the venture created a genuinely new product that no member could offer alone, so the rule of reason applied.

Texaco Inc. v. Dagher, 547 U.S. 1 (2006) went further: where two oil companies formed a lawful production and marketing joint venture, the venture's pricing of its own products is not price fixing at all — a firm setting the price of what it sells is engaged in core business conduct, not a horizontal restraint. Dagher is the strongest authority available to a properly integrated venture.

But NCAA v. Board of Regents of the University of Oklahoma, 468 U.S. 85 (1984) shows the limit: a venture that restrains output and price beyond what the integration requires loses. The restraint must be reasonably necessary to the venture's legitimate purpose.

Practical compliance architecture

Define the venture's scope narrowly and in writing. The narrower the field, the easier to defend restraints within it.
Limit information flow. Competitively sensitive information — current pricing, costs, customer-specific terms, forward plans outside the venture's field — should not move between parents through the venture. Use clean teams and an information protocol.
Draft non-competes to the field. A non-compete covering the venture's actual business for the venture's duration is generally ancillary and defensible. One covering adjacent markets or extending long after exit is not.
Screen for HSR. Formation of a venture with contributions above the thresholds is a reportable acquisition. Do the analysis at term sheet stage, not at signing.
Watch 15 U.S.C. § 2 where the parties have substantial shares; a venture can be a vehicle for monopolization or attempted monopolization independent of § 1.

Deadlock and exit: the provisions that decide who owns the business

Escalation first

Every venture agreement should require escalation before anything drastic: the matter goes to designated executives at each parent, then to the chief executives, with defined time periods. A surprising share of deadlocks resolve here, because the operating people who deadlocked report to people with a broader view.

Mediation is a sensible next step. Arbitration of a business deadlock is generally not — arbitrators are good at deciding legal disputes and bad at deciding whether to enter Brazil.

The buy-sell mechanisms

If escalation fails, one party must be able to acquire the other's interest. The mechanisms differ in who sets the price and who chooses.

Shotgun (Texas shootout). Party A names a price per unit; Party B must either buy A's interest at that price or sell its own at that price. Elegant, self-policing on price — and systematically favors the party with more cash and better information. A partner that cannot finance a purchase must sell at whatever price the other names. Use only where the parties are comparable in size and sophistication, or add a floor.

Modified shotgun. Variations that reduce the asymmetry: a right of first offer with a fairness floor; a requirement that the triggering party fund into escrow; a valuation collar; or a rule that only the non-triggering party may initiate for a period.

Put and call rights. A partner may put its interest to the other (or the venture) at a formula or appraised price, or the other may call it. Cleaner than a shotgun and easier to finance because the price is knowable in advance. The design question is the price standard — fixed multiple, formula on EBITDA, appraised fair market value, or a ladder that changes over time.

Right of first refusal and right of first offer. ROFR lets a partner match a third-party offer; ROFO requires the seller to offer to the partner first. ROFR depresses value because third parties will not spend diligence money on a deal a partner can take away at the last moment. ROFO is generally better for the seller and nearly as protective for the holder.

Forced sale / drag. After a stated period or on deadlock, either party may cause a sale of the whole venture to a third party, with both dragged. This is the only mechanism that reliably produces a market price, and it is the right default for ventures where neither party is certain to be the long-term owner.

Dissolution. The nuclear option. Available by default in many agreements and almost never a good outcome, because a venture's value is usually in the going concern rather than the assets. Consider requiring dissolution to be a last resort after a forced-sale process fails.

The valuation mechanism

Whatever the exit, someone must set a price. Options:

Formula (a multiple of trailing EBITDA, adjusted for debt and cash). Predictable, cheap, and wrong whenever the business changes character.
Appraisal. Each side appoints an appraiser; if they differ by more than a stated percentage, a third decides or averages. Slower and more expensive, but robust.
Baseball appraisal. Each side submits a number and a neutral picks one. Powerfully disciplines both sides toward reasonableness.
Market test. Solicit third-party bids and use the best as the price. The most accurate and the most disruptive.

Whatever mechanism is chosen, address minority and marketability discounts explicitly. A buy-sell that says "fair market value" without specifying whether the interest is valued as a pro rata share of the whole or as a minority block invites a dispute worth tens of millions.

Economics: distributions, tax, and the things that quietly diverge

Ownership percentages get all the attention in a term sheet, and they are usually the least consequential number in the document. What determines who gets what is the distribution waterfall, the tax allocation, and the affiliate-transaction terms.

The distribution waterfall

Even a 50/50 venture rarely distributes 50/50 forever. Common layers:

Tax distributions — mandatory distributions sufficient to cover members' tax liability on allocated income. Without this, a member can owe tax on phantom income while the venture reinvests. Every pass-through venture needs a tax distribution provision, and it should be a first-priority, non-discretionary payment computed at an assumed rate.
Preferred return on unreturned capital — where one member funded more, or funded later, a preferential return compensates for the timing and risk difference.
Return of capital, in a specified order.
Residual sharing, by percentage interest.

The negotiation lives in layer two. A member who funds a disproportionate share of growth capital wants a preference; a member contributing non-cash value resists one, because a preference converts the other's cash into a senior claim.

Tax allocation and the special problems of contributed property

When a member contributes appreciated property, the difference between the property's tax basis and its agreed value creates built-in gain that must be allocated back to the contributing member when the property is sold or depreciated. This is not optional and it is not intuitive, and ignoring it produces an unpleasant surprise years later.

Practical consequences:

The contributing member bears the tax on the pre-contribution appreciation, which means its after-tax economics differ from the other's even at identical percentage interests.
Depreciation allocations shift, so book and tax capital accounts diverge and must be tracked separately.
Contributed intellectual property raises additional questions about whether the transfer is a contribution, a license, or a sale, with sharply different treatment.

Advice: bring tax counsel in at the term sheet, not at signing. The choice between contributing an asset and licensing it, or between contributing property and selling it to the venture for cash and equity, often changes the deal's value by more than the percentage split being argued about.

Affiliate transactions

Ventures buy from and sell to their parents constantly — raw materials, shared services, distribution, IT, insurance, real estate. Each of those is a channel through which value can move out of the venture and into one parent.

The controls that work:

Arm's-length pricing standards with a defined methodology (cost plus a stated margin, or benchmarked to third-party terms);
Approval thresholds requiring the non-transacting member's consent above a dollar amount;
Audit rights over the affiliate's cost records supporting charged amounts;
Termination rights on the venture's side, so it is not locked into a parent's services indefinitely; and
A benchmarking mechanism that periodically tests shared-service pricing against the market.

The most common leakage is not fraud; it is a parent charging its fully loaded corporate overhead allocation to a venture that consumes a fraction of it. Specify what may be allocated and what may not.

Minority protections in an unequal venture

Not all ventures are 50/50. Where one party holds 70% or 80%, the minority's entire position rests on negotiated protections, because default entity law gives it very little — particularly in an LLC where duties have been waived.

The essential minority package:

Reserved matters, as above, plus specific protection against issuing equity at below fair value, amending the agreement in a way that disproportionately affects the minority, and changing the distribution waterfall.
Anti-dilution, either a preemptive right to participate pro rata in new issuances or a price-based adjustment.
Information rights: audited annual financials, quarterly management accounts, the annual budget, board materials, and access to management. Do not rely on statutory inspection rights; an LLC agreement can restrict them, and many do.
Tag-along rights, so a majority sale does not strand the minority in a venture with a new and unknown partner.
A put right, exercisable on defined triggers — change of control of the majority, a material breach, failure to distribute for a period, or simply after a stated number of years. The put is the minority's single most valuable protection, because it converts an illiquid position into a claim for money.
Deadlock-independent exit, since a minority cannot create a deadlock and therefore cannot use a deadlock-triggered buy-sell.

A drafting note on puts. A put is worth nothing if the majority cannot pay it. Consider a parent guarantee, a payment schedule with interest, a security interest in the majority's units, or a fallback right to force a sale of the venture if the put is not honored.

Change of control of a partner

Ventures outlive the strategies that created them, and partners get acquired. A venture agreement that does not address a partner's change of control has left its most predictable crisis unmanaged.

Consider the range of outcomes when Partner A is acquired by Partner B's largest competitor:

Nothing happens, because the agreement is silent. The venture now shares confidential information with a competitor through A's appointed managers, and B's information protocol is worthless.
The agreement gives B a call right at a formula price. B acquires the venture, possibly at a discount if the formula is punitive.
The agreement gives B a put right, allowing exit at fair value.
The agreement triggers a forced sale of the whole venture.
The agreement suspends A's governance and information rights pending a resolution period — a useful intermediate step that preserves value while the parties negotiate.

Define the trigger carefully. "Change of control" should capture acquisition by a competitor specifically, not merely any change in ownership; a private equity recapitalization of a partner is a different event from acquisition by a rival. Many agreements use a two-tier trigger: broad consequences for a competitor acquisition, narrow ones otherwise.

Also address the reverse: what happens when a partner's parent simply loses interest, cuts the venture's staffing, or stops performing its commitments without any formal change of control. That is the more common failure, and the remedy is the commitment covenants and dilution mechanics discussed above.

Worked example: the Brightwater venture

The parties. Corriveau Systems makes industrial water-treatment hardware. Ndiaye Analytics builds sensor and software platforms. They form Brightwater LLC to sell an integrated monitoring-plus-treatment offering to municipal utilities — a product neither can offer alone.

Contributions. Corriveau contributes $40 million cash and a manufacturing line valued at $20 million. Ndiaye contributes a field-limited, royalty-free, perpetual license to its sensor platform for water-treatment applications, valued at $60 million, plus twelve seconded engineers. Ownership is 50/50.

The first problem — the license. Ndiaye's license is limited to "water-treatment applications" and terminates if Ndiaye exits the venture. Corriveau's counsel, Ilse Brandão, catches both points. If the license dies on Ndiaye's exit, then Ndiaye holds a permanent option to destroy the venture's value, which means Ndiaye's interest is worth far more than 50% and Corriveau's far less.

The fix, negotiated: the license survives any exit, becomes irrevocable on the venture's second anniversary, and expands automatically to cover improvements Ndiaye makes in the field for five years. In exchange, Ndiaye receives a royalty on venture sales after a volume threshold and a narrower field definition. The economics moved from the ownership split into the license terms, which is usually where they belong.

Governance. Four managers, two per side. Reserved matters require unanimity. The annual budget is a reserved matter — but with a deemed-budget fallback at prior year plus 4%, with capital expenditure capped at the prior year's level.

Year three — divergence. Brightwater is profitable. Corriveau wants to reinvest in a European expansion requiring $70 million of new capital. Ndiaye, whose parent is under earnings pressure, wants distributions and refuses to fund.

The deadlock provision routes to the chief executives, who fail to agree in sixty days. Now the mechanics matter.

The capital call provision permits Corriveau to fund the shortfall as a member loan at SOFR plus 700 basis points, convertible after eighteen months into equity at 85% of appraised value. Corriveau funds. Ndiaye's stake begins to shrink in expectation.
Ndiaye's response is to invoke the buy-sell. The agreement uses a baseball appraisal put/call: on deadlock, either party may trigger, and each submits a value for 100% of the venture; a neutral appraiser selects one; the triggering party then chooses whether to buy at that value or sell at it.

Ndiaye submits $520 million. Corriveau submits $610 million. The neutral selects Corriveau's number, reasoning that the European pipeline is real. Ndiaye, as trigger, must now choose: buy Corriveau out for $305 million, or sell for the same. Ndiaye's parent will not fund $305 million. Ndiaye sells.

What the drafting did. Notice that Corriveau won, and won because of provisions negotiated in year zero: a convertible member loan that made non-funding expensive, a deemed budget that prevented paralysis, and a baseball mechanism that punished Ndiaye's low submission. None of these were the "important" terms in the term sheet; the term sheet fight was about the ownership split, which turned out not to matter.

What Ndiaye should have negotiated. A cap on mandatory dilution; a right to convert to a non-diluting preferred interest on non-funding; a requirement that the triggering party fund a deposit into escrow; and a floor on the buy-sell price tied to an independent valuation. Any one of them changes the outcome.

Cross-border ventures: the additional layer

International ventures add problems that domestic drafting does not anticipate.

Local ownership and licensing requirements. Many jurisdictions restrict foreign ownership in specific sectors — telecommunications, defense, media, natural resources, financial services — or require a local partner above a stated percentage. The local-partner requirement is frequently the reason for the venture, which means the foreign partner's leverage is structurally weak and must be rebuilt contractually through reserved matters, service agreements, and IP licensing rather than through equity.

Governing law and dispute resolution. For a cross-border venture, neither partner will accept the other's home courts, and the entity's own jurisdiction may be neither. The standard solution is arbitration under institutional rules, seated in a neutral jurisdiction, with the New York Convention providing enforcement. Specify the seat, the institution, the number of arbitrators, the language, and — importantly — whether emergency relief is available, since a venture dispute often needs an injunction before a tribunal exists.

Currency, repatriation, and dividends. Exchange controls can trap distributions in the venture's jurisdiction. Address the currency of capital contributions, the currency of distributions, who bears exchange risk, and what happens if repatriation becomes unlawful or impracticable.

Anti-corruption. A venture in a high-risk jurisdiction, particularly one with a local partner who provides "government relations," is the classic Foreign Corrupt Practices Act exposure. The foreign partner needs audit rights, compliance covenants, training obligations, termination rights for compliance breaches, and — critically — actual visibility into the venture's third-party intermediaries. A compliance clause without audit rights is decorative.

Tax treaty and structuring. The choice of holding jurisdiction affects withholding on dividends, interest, and royalties, and the availability of treaty benefits. Structures that were routine a decade ago may now fail principal-purpose tests. This is specialist work and should be done before the entity is formed, because migrating a venture later is expensive and sometimes taxable.

Employment and secondment. Seconding employees into a foreign venture can create permanent establishment exposure for the seconding parent, immigration obligations, and local employment protections that attach to the individual regardless of the contract's terms.

Antitrust information protocols in practice

Because the information question recurs in every competitor venture, it is worth stating the operational answer rather than the doctrinal one.

Build three tiers.

Venture-only information. Data generated by and about the venture's own business. Flows freely within the venture; flows to parents only as needed for financial reporting, tax, and governance, and then in aggregated form where possible.
Parent information the venture needs. Technical specifications, cost inputs required for the venture's products, regulatory data. Flows in, subject to confidentiality and use restrictions.
Competitively sensitive parent information. Current and future pricing, customer-specific terms, margins, strategic plans, and capacity plans outside the venture's field. Does not flow at all, in either direction, and does not sit on shared systems.

Implement it with:

A written information protocol adopted at formation and referenced in the venture agreement;
Segregated systems and access controls, not merely policies;
Clean-team arrangements for any diligence or integration work;
Training for seconded employees, who are the most likely vector because they retain their parent's email and habits;
Counsel attendance at board meetings where scope or pricing is discussed; and
Minutes that record the antitrust framing of decisions about output, price, and territory.

Why it matters beyond compliance: in litigation, the existence of a documented protocol is the difference between a venture that looks like an integration and one that looks like a cartel with a governance document. The protocol is evidence, and its absence is evidence too.

Termination, unwind, and the things nobody plans

When a venture ends, several problems arrive simultaneously:

Licensed IP reverts or survives, per the agreement. Improvements made by the venture need an owner, and the answer should be in the document.
Employees — seconded people return; venture-hired employees need a home or a severance plan.
Customer contracts need assignment, and many contain change-of-control provisions triggered by the unwind.
Shared services from a parent must continue during a transition, which means a transition services arrangement negotiated at the worst possible moment unless one was pre-agreed.
Non-competes either bind the exiting party or do not; the duration and scope should be set at formation, when neither party knows which side of the clause it will be on.
Regulatory licenses and permits held by the venture may not be transferable.
Tax consequences of a liquidating distribution differ sharply from those of a sale of interests.

The single best practice is to draft a short "unwind protocol" at formation: a schedule listing which assets go where, which agreements survive, and how long transition services run. It takes a day to negotiate at formation and is impossible to negotiate at the end.

Ten questions that predict whether a venture will work

Ask these before drafting. If the parties cannot answer them together, the venture is not ready regardless of how good the term sheet looks.

What can this venture do that neither party can do alone? If the honest answer is "nothing, but it splits the cost," consider a contract instead.
Who is the chief executive, and to whom does that person actually report? A venture CEO with two bosses has none.
What happens in year four when one partner wants to reinvest and the other wants cash? This is the most common divergence, and the answer is the distribution and capital call provisions.
If the venture needs $100 million, where does it come from? Pro rata, third-party debt, or one partner — and what happens if the other cannot match.
What does each partner's contributed IP look like in ten years, and who owns the improvements?
Can either partner compete with the venture, and where is the boundary? Write the field definition; do not gesture at it.
What information cannot move between the parents, and how is that enforced technically?
How does a partner get out, and can the other actually afford to buy? Price the exit at formation.
What happens if one partner is acquired by the other's competitor?
Who unwinds this, and what does the venture's customer contract say about assignment?

The organizing principle. A joint venture agreement is not a document about how two companies will cooperate. It is a document about what happens when they stop. Draft it in that spirit, and the cooperation tends to take care of itself.