Most executives assume a partnership falls apart in execution — misaligned teams, poor integration, slow go-to-market. While that is not incorrect, after structuring and closing more than $2.3 billion in partnerships and joint ventures across Japan, ASEAN, and the US, I can tell you the opposite is often true.
Failure can be structural — designed in at the term sheet stage, ratified in the negotiation, and sealed at signing. By the time the problems arise, the architecture that produced them — and could otherwise have fixed them — is already agreed.
This is not an execution failure. It is a deal design failure — and it is far more common than the post-mortems ever admit.
THE ANATOMY OF A PRE-ORDAINED FAILURE
When I was the lead architect of Amazon Prime’s telco partnership program in Japan — closing $950 million in combined deal value across NTT Docomo and KDDI — both sides were deliberate. The deal teams were professional. The timelines were measured. The governance was designed carefully, not deferred.
I have spent the years since working across a range of mandates — including deals that succeeded and deals that didn’t. The contrast is instructive. The Amazon Prime telco program worked in part because the governance was deliberate, the incentive economics were data-driven, and both sides remained aligned on what success meant. HOOQ is a different lesson. When new learning emerged — as it always does — there was no governance architecture to absorb it. Goals that were never fully aligned at the pre-deal stage diverged further as the market moved. Without a mechanism to adapt, the partnership couldn’t protect itself from what it was discovering. The blind spots that produced that outcome are consistent enough across mandates that I now use them as a diagnostic before any term sheet is issued.
Not in legal. Not in IP or exclusivity or revenue-share mechanics — those have advocates on both sides of the table. The blind spots are in the economic gearing of KPIs and in governance. In decision rights. In the question nobody wants to ask out loud: what happens when the commercial model does not perform as modeled in year two? What happens when the strategic rationale that closed the deal no longer reflects either party’s priorities?
Asking those questions signals doubt — and doubt is socially awkward when everyone in the room wants the deal to work. So they get noted, deferred, and quietly dropped from the agenda.
So the governance gets written loosely. The KPIs get defined aspirationally rather than operationally. The dispute resolution clause gets copied from a template. And two years later, someone is sitting across a table from a partner they no longer trust, arguing over what the contract meant.
THE FOUR STRUCTURAL FAILURE MODES
Across my career in partnerships, failure clusters around four root causes. They are not equally distributed across industries or geographies, but they appear consistently enough that I use them as a diagnostic before any term sheet is issued.
Incentive inversion. The commercial model pays one party for behavior that disadvantages the other. This is most common in distribution and content partnerships where one party is a reseller: the reseller’s incentive is margin maximisation, not partner brand stewardship. When the incentive structure is never explicitly addressed in the deal design, the behavior that follows is not a betrayal — it is entirely rational. HOOQ, the $92 million joint venture conceived from within SingTel with Warner Bros. and Sony Pictures as partners, is an example of what happens when goals are not sufficiently aligned at the pre-deal stage and when no governance mechanism exists to protect the partnership as new learning emerges. The equity structure was intended to force a shared upside frame — but without the governance architecture to sustain alignment as the market evolved, good intentions were not enough.
Governance vacuum. This is the failure mode most people mean when they say a partnership broke down — but they usually locate the cause in the wrong place. Governance fails not because the governance structure was absent, but because the economic architecture underneath it was never properly specified. The KPI-to-waterfall relationship — what gets measured, how it is measured, what thresholds trigger which commercial consequences — sets the performance boundaries that governance is supposed to enforce. Without that specification, governance has nothing to act on. The legal entity exists. The committee is specified. But when the first commercial dispute arises — and it nearly always does — there is no agreed mechanism to resolve it, because nobody defined what resolution was supposed to look like. It escalates to principals who are busy and increasingly irritated. The relationship cools. The partnership drifts. This is not a people problem. It is an architecture problem.
Motivational decay. Partnerships are typically engineered around outcomes — the signing, the milestone, the revenue target. What rarely gets designed is the journey between them. At month four, the novelty has worn off. The internal champion who closed the deal has moved to another role. Both organisations are competing for resources against other priorities. The partnership is contractually alive but operationally drifting.
This is motivational decay — and it is almost never addressed in the deal architecture because it doesn’t feel like a deal question. That miscategorisation is the problem. By the time it’s recognised as one, the deal is already signed and the leverage to fix it is gone.
The remedy is not better management. It is intermediate incentive design — milestone structures that reward progress, not just completion; joint early wins built deliberately into the operating plan to generate shared momentum; governance cadences that surface and celebrate what’s working, not just escalate what isn’t. When one party feels the economics are extractive rather than generative — that they are paying tolls rather than sharing upside — decay accelerates. Discretionary effort disappears. Both sides fulfil the minimum contractual obligation and nothing more. You cannot govern your way out of that. It has to be designed out before signing.
The partnerships that sustain are the ones where both parties remain commercially motivated at month fourteen, not just month one. That motivation doesn’t happen by itself. It has to be engineered.
Information asymmetry baked into the structure. One party has superior visibility into the metrics that determine payouts, milestone triggers, or renewal decisions. In AI partnerships, this is now a critical issue. If your data partner controls the measurement environment, they also control the evidence base for your commercial entitlements. I have seen this pattern repeatedly across AI commercialisation programs — and its consequences go beyond disputes. When measurement custody is unclear or contested, outcome-linked compensation structures become impossible to enforce, so nobody proposes them. Deals default to blunter instruments — fixed fees, flat revenue share — that are easier to administer but leave value unrealised on both sides. Innovative commercial structures require trustworthy measurement. That is why I now treat KPI measurement custody as a first-order commercial term, not an afterthought resolved after close.
WHAT STRUCTURAL DEAL DESIGN ACTUALLY LOOKS LIKE
The antidote to each of these failure modes is not more lawyers or longer contracts. It is earlier commercial architecture work — before the term sheet is issued, not after it is drafted.
This means building a model that maps the incentive structure explicitly, stress-testing it under downside scenarios, and asking: does each party’s rational behavior in this model produce the outcome the partnership requires? If not, the commercial model needs to change before the lawyers enter the room.
It means defining governance at two levels — the board-level governance that manages strategic alignment, and the operating-level governance that manages weekly execution — and specifying decision-right thresholds that keep routine decisions out of the principal layer entirely. Critically, and this is a principle I argue at length in The New Rules of Partnerships, it means specifying the KPI-to-waterfall architecture before governance is written, so that governance has enforceable performance boundaries to work from rather than aspirational ones to interpret.
And it means, in an era where AI and data are central to commercial outcomes, insisting on neutral measurement custody. This is the partnership equivalent of an independent audit function, and it is one of the deal design requirements the book treats as non-negotiable. You do not negotiate a revenue-share arrangement without independent account reconciliation. You should not negotiate an AI-powered partnership without independent KPI attestation.
The sequencing matters as much as the substance. This work belongs in the commercial architecture phase — before legal drafts begin, before the term sheet is issued, and before both parties have spent enough political capital on the deal to make renegotiating any term feel like failure.
THE DISCIPLINE THAT CHANGES THE OUTCOME
The executives who close partnerships that deliver are not doing something exotic. They are doing the structural work earlier than their peers, with more rigour, and with a clearer model of how the deal actually functions under operational stress.
That discipline — what I call dealcraft — is not a methodology you read about. It is a practice built from closed deals, survived failures, and the specific pattern recognition that only comes from operating at this level across multiple markets and technology cycles.
In the AI era the stakes are higher and the structures are more complex. Data provenance, KPI measurement custody, dual-track governance for cross-border JVs — these are not standard legal terms. They are architectural decisions that most internal partnership teams have never had to make, because deals at this level of complexity are relatively new.
The executives who get this right either built this capability over decades of their own experience, or they brought in someone who had. The window to make that call is before the term sheet — not after the contract is signed and the failure mode is already built in.
Randy McGraw is the founder of M2 Ventures and the author of The New Rules of Partnerships. He has been the lead day-to-day architect of more than $2.3 billion in commercial partnerships and joint ventures across Japan, ASEAN, and the US — including Amazon Prime’s telco deals with NTT Docomo and KDDI, and the HOOQ joint venture with Warner Bros. and Sony Pictures at SingTel. He is a Managing Partner at Andersen Consulting. M2 Ventures advises corporate clients on partnership architecture, negotiation, and governance.