Every essay in this series has argued that cooperating with rivals can grow pies no single firm could bake. This one is about the condition that makes all of it survivable: the fence. Cooperation transfers things, information, access, exposure, and some of what your firm holds must never transfer, because it is the source of your added value itself. Call these the crown jewels: the customer relationships, the recipes, the data, the pricing logic, the supplier terms that make your disappearance expensive. A partnership that leaks them converts your irreplaceability into the partner’s education, and the asymmetric learning trap closes quietly: the rival who once needed you, no longer does. Ring-fencing is not paranoia and it is not bad faith. It is the discipline that lets good faith operate safely, the lock that makes the open door possible. Generous on the floor, guarded in the vault: both, deliberately, at once.
Key Takeaways
- Crown jewels are the assets your added value lives in: whatever makes your exit expensive for others. Everything else is furniture, and furniture can be shared.
- The four-question audit identifies them: what would a rival need to replace you; what took years to build; what do customers actually stay for; what would you never show a partner’s analyst?
- The four fences: scope fences (what the cooperation may touch), people fences (who attends what), artifact fences (what documents and systems partners see), and time fences (what expires when the partnership does).
- Fencing failures are mundane, not cinematic: the shared dashboard with too many tabs, the plant tour that included the process line, the joint bid that revealed the cost model.
- Over-fencing kills cooperation too: partners who share nothing add nothing. The fence exists to make generosity safe, not to replace it.
- Fences are declared, not hidden: naming what is out of scope builds more trust than pretending everything is open.
What actually counts as a crown jewel?
Less than founders fear and more than they protect. The test is the added-value question inverted: which specific assets, if a rival absorbed them, would erase what the market pays you for? Four questions surface them. What would a competitor need to learn or copy to make customers indifferent between you and them? What took years, relationships, or failures to build that money alone cannot shortcut? What do your customers actually stay for, not what you advertise, but what would make them leave if it degraded? And the instinct test: what would you quietly refuse to show a partner’s bright young analyst on a site visit?
Run those four and the usual suspects sort themselves. Customer lists and relationship history: jewels, almost always. The pricing model and its logic: jewels. Supplier terms and sourcing routes that took a decade: jewels. Proprietary data, the repayment histories, the demand patterns, the formulation: jewels. Meanwhile the things firms instinctively hide often are not: the org chart, the general process everyone in the industry runs, the market information any diligent observer could assemble. Guarding furniture while the vault stands open is the standard failure, because furniture is visible and jewels are ambient, they leak through conversation, dashboards, and site visits, not through burglary.
How do the four fences work?
Scope fences define what the cooperation may touch, in writing, at formation. The consortium bid covers this tender; the shared cold store covers storage and nothing upstream; the coalition covers regulatory voice, not commercial data. Scope creep is the fence’s natural enemy, cooperation drifts toward whatever is useful, so the fence needs a standing review: is anything now flowing that the founding scope excluded?
People fences control who attends what. The partnership needs your operations manager; it does not need your head of sales socializing quarterly with theirs. Rotate liaison roles so no single employee becomes the partner’s open window; brief the liaisons explicitly on what is fenced, because employees default to helpfulness, and helpfulness is how jewels walk.
Artifact fences govern documents and systems: partners see outputs, not engines. The joint venture gets the delivery schedule, not the routing algorithm; the co-bidder gets your price, not your cost build-up; the shared dashboard shows the agreed metrics and only those. Most real leaks are artifact leaks, the spreadsheet with the extra tab, the system login scoped too wide, and the remedy is preparing partner-facing versions of things rather than sharing working files.
Time fences decide what survives the partnership’s end: return-or-destroy clauses for shared data, non-solicitation of each other’s staff and customers for a defined period, and honest recognition that knowledge cannot be un-learned, which is why the deepest jewels are protected by never transferring, not by contracts about after.
Where is the line between prudence and paranoia?
At declared generosity. The fence fails in two directions, and the second is under-preached: partners who share nothing add nothing, and a cooperation of mutual vaults produces mutual nothing. The pie only grows when real assets meet across the table; fencing exists to make that meeting safe, not to prevent it. The operators who do this well are strikingly open about the fence itself: “our customer data and cost model are out of scope, everything else is on the table” builds more trust than vague openness that partners later discover had invisible walls. Declared fences are respected; discovered ones are resented.
The posture, once more, is the layered rival’s: whole-hearted on the shared layer, locked on the core, and never confusing the two. For the faith-driven operator the theology is oddly parallel: wisdom texts praise both the open hand and the guarded heart, generosity as practice and prudence as duty, and treat neither as contradicting the other. Guard what was entrusted to you to steward; give freely what was given to be shared; and know, in writing, which is which. That written knowledge, kept current as the partnership deepens through each notch of structure, is the whole art of cooperating with someone who could, one day, be across the table in a harder conversation. Fence well, and that day never has to come; fence nothing, and it is already scheduled.
FAQ
What are crown jewels in a business partnership?
The assets your market power actually lives in: customer relationships and history, pricing logic, proprietary data, hard-won supplier terms, and any capability that took years to build. The test: what would a rival need to absorb to make customers indifferent between you?
What are the four fences?
Scope fences (what the cooperation may touch, written at formation), people fences (who attends what, with rotating liaisons), artifact fences (partners see outputs, never engines), and time fences (what expires or returns when the partnership ends).
How do crown jewels usually leak?
Mundanely: shared dashboards with extra tabs, over-scoped system access, plant tours that included the process line, helpful employees answering one question too many. Preparing partner-facing versions of documents prevents most of it.
Can you over-fence a partnership?
Yes, fatally: partners who share nothing add nothing. The fence exists to make generosity safe. Declare what is out of scope openly and be genuinely open with the rest; declared fences build trust, discovered ones destroy it.
Do contracts protect against a partner learning your business?
Only partially. Non-disclosure and non-solicitation clauses deter and compensate, but knowledge cannot be un-learned. The deepest jewels are protected by never transferring them, which is what the artifact and people fences are for.
Related Reading
- Complementary vs Substitutive: The Test That Settles It
- Added Value Is the Real Currency
- The Spectrum of Structures: From Handshake to Joint Venture
- Co-opetition Is Not a Contradiction
Sources and Evidence
- Brandenburger and Nalebuff, “The Right Game: Use Game Theory to Shape Strategy,” Harvard Business Review (1995): protecting added value while cooperating.
- Co-opetition (Brandenburger and Nalebuff, 1996), overview: information and learning risks in cooperative-competitive relationships.