Two companies sign a partnership. There is a joint announcement, a handshake photo, mutual logos on each other's websites, and twelve months later, not one dollar of revenue can be traced to it. This is the default outcome, and it is not because the idea was wrong. It is because nobody treated the partnership as a channel that has to be built, resourced and measured like any other.

The quick version

  • A partnership is a revenue channel, not an event. Treat it like one: an owner, a number, a pipeline, a forecast, the same machinery you give direct sales.
  • The economics are real but unforgiving. Ernst and Bamford report that the overall success rate of alliances hovers near 50%, and that a typical large company draws roughly 15–20% of its revenue from them, so the upside is big and the failure rate is high.
  • Cross-sector ecosystems are becoming a structural channel, not a fringe one: McKinsey's "sectors without borders" research, the work behind Venkat Atluri and Miklós Dietz's The Ecosystem Economy, puts roughly 30% of global revenue, on the order of $60 trillion, inside cross-sector ecosystems by 2025.
  • The move is to pick the right type of partnership for the outcome you want, fund the joint motion (not just the contract), and gate the relationship on value delivered to the end customer, not on the press release.

The idea in depth

"Partnership" is one of the most overloaded words in commercial life. A referral arrangement, a reseller agreement, a technology integration and a fifty-fifty joint venture are all called partnerships, and they behave nothing alike. Before you can run partnerships as a channel, you have to decide which kind of partnership you are actually building, because that decides who sells, who delivers, and where the revenue lands.

flowchart TD
    P(["A 'partnership'
(decide which kind)"]) --> R(["Referral / affiliate:
they point, you sell"])
    P --> C(["Channel / reseller:
they sell your product"])
    P --> T(["Technology / integration:
you fit into their product"])
    P --> A(["Strategic alliance / JV:
you build & go to market together"])
    R --> O(["Incremental revenue
you couldn't reach alone"])
    C --> O
    T --> O
    A --> O
Four common partnership types, ordered by depth of commitment, each is a different channel with different owners, economics and risks. Leaders Loop

Why partnerships are a channel, and why most are run as an event

The strategic case is now hard to wave away. McKinsey senior partners Venkat Atluri and Miklós Dietz, whose work on the ecosystem economy they discuss in McKinsey's "Author Talks", build on the firm's "sectors without borders" research: it estimated that around a dozen cross-sector ecosystems, networks of organisations selling across traditional industry borders, could account for roughly 30% of global revenue, on the order of $60 trillion, by 2025. Translate that out of the consulting register and it says something plain: more and more of what customers buy will reach them through someone else's relationship, not through your own salesforce. A company with no partner motion is opting out of a growing share of the market.

And yet the failure rate is stubborn. Reviewing decades of alliance practice in their Harvard Business Review article "Your Alliances Are Too Stable" (2005), David Ernst and James Bamford note that the overall success rate of alliances hovers near 50%, and that a 2004 McKinsey survey found at least 70% of companies' major alliances were underperforming and needed restructuring. The same article observes that a typical corporation depends on alliances for something like 15–20% of its revenue, assets or income. So this is not a sideshow that occasionally misfires; it is a material chunk of the business that fails roughly half the time.

The reason for the failure rate is almost always the same: the partnership was treated as a signing rather than a channel. A contract gets negotiated, a launch gets announced, and then the thing is left to run itself, with no quota, no pipeline, no joint account planning, and no one whose job depends on it producing revenue.

So resource the partnership like a channel from day one. Give it a named owner with a number, a pipeline you can review weekly, a forecast that rolls into the company's, and a clear answer to "what does each side actually do to make a sale happen?" If the only artefact your partnership produces is a logo on a webpage, you have built an event, not a channel.

The classic trap: a partnership where the value runs one way

A durable partnership has to pay both sides, repeatedly. When it doesn't, it quietly dies, or worse, one party is effectively buying access while the other coasts. Ernst's earlier work with Joel Bleeke ("Is Your Strategic Alliance Really a Sale?", HBR 1995) made the sharp point that many "alliances" are really disguised acquisitions or one-sided dependencies dressed in the language of collaboration, and they behave accordingly once the initial enthusiasm fades.

A partnership that only pays one side isn't a channel, it's a subsidy with a launch event.

The practical version of this trap shows up in channel and reseller deals. You recruit a hundred partners, none of them sells anything, and you conclude "partnerships don't work for us." Usually the truth is narrower: those partners had no reason to prioritise your product over the dozen others in their bag, because there was nothing in it for them beyond a margin they could earn more easily elsewhere. Jay McBain, a long-time analyst of the partner ecosystem, frames the modern shift in "Halftime in the Decade of the Ecosystem" as a move away from transactional reselling toward partners who influence a buyer's decision long before a vendor's salesperson shows up. The partners worth having are the ones your customers already trust.

Design the partner's win before you design your own. For every partnership, write one sentence: "This partner makes more money / serves their customer better / wins more deals because they work with us." If you can't finish that sentence honestly, the partnership will underperform no matter how good the contract looks. (This is the same logic as upsell, cross-sell and land-and-expand, expansion only compounds when value genuinely flows; a partnership is the same principle pointed outward.)

The honest limitation: partnerships are slow, low-control, and not free

Here is where the enthusiasm needs a cold towel. A partner channel is not a shortcut to cheap revenue. It is slower to stand up than hiring a salesperson, you control far less of the customer experience, and "co-selling" frequently means two companies politely waiting for the other to do the work. The 50% success rate Ernst and Bamford cite is a real floor, not a worst case, and joint ventures, the deepest form, had an average lifespan in their data of just five to seven years.

There are also failure modes specific to the channel. Channel conflict, your direct salesforce and your partners chasing the same deal, can poison both motions. Partner-sourced revenue is harder to forecast because you don't own the pipeline. And the ecosystem and revenue figures above are directional estimates and benchmarks, not a promise: the $60-trillion ecosystem number is a projection about a whole economy, not your funnel, and a 15–20% revenue share is a typical figure across large firms, not a target you should expect from year one. Treat all of it as a case for building the muscle, not as a forecast you can bank.

The fix is to start narrow and prove the motion before you scale it. Pick one or two partners, one clear joint offer, and one measurable outcome (sourced pipeline, influenced deals, time-to-value for a shared customer). Make that single relationship genuinely work, repeatable, profitable for both sides, forecastable, before recruiting the next ten. A channel you can't yet measure is a channel you can't yet scale.

A worked example

The figures below are illustrative, chosen to show the mechanics rather than to report a real company.

Imagine a mid-size payroll software company that sells direct to small businesses. Growth from its own salesforce is getting expensive, each new customer costs more to acquire than the last. Its leaders decide to test partnerships as a channel rather than buy more ad inventory.

They resist the instinct to recruit fifty "partners" at a launch event. Instead they pick one: a regional accounting firm whose clients are exactly the small businesses the software serves, and who are constantly asked by those clients "which payroll system should I use?" The win for each side is written down first. The accounting firm earns a referral fee and, more importantly, looks good to clients by recommending a tool that makes their own bookkeeping cleaner. The software company reaches buyers at the precise moment they're asking for a recommendation, from a source those buyers already trust.

Then they build the motion, not just the contract: a named partner manager on the software side, a simple shared referral link, a quarterly joint review, and one co-branded "payroll for small business" webinar. The relationship is measured on sourced pipeline and on how fast referred customers get up and running.

flowchart LR
    S(["Direct sales only:
CAC rising each quarter"]) --> Q(["Pick ONE partner:
trusted accounting firm"])
    Q --> W(["Write each side's win
before signing"])
    W --> M(["Build the motion:
owner + referral link
+ joint webinar"])
    M --> V(["Measure: sourced pipeline
& time-to-value"])
    V --> E(["Proven, profitable
both sides → recruit
the next partner"])
One partner, proven end to end, before scaling the channel, the opposite of the fifty-logo launch. Illustrative example. Leaders Loop

Six months in, the single partner is sourcing a steady trickle of well-qualified customers who close faster and stick longer, because they arrived pre-sold by someone they trust. Now, and only now, does the company have something worth copying: a repeatable, two-sided, measurable motion it can take to the next twenty accounting firms. Had it launched with fifty partners and no motion, it would have learned nothing except that "partnerships don't work."

Frequently asked questions

What's the difference between a partnership, an alliance and a channel?

Loosely: a channel partner (reseller, distributor) sells your product on your behalf; a referral partner points customers your way but you do the selling; a technology partner integrates with you so the combined product is more valuable; and a strategic alliance or joint venture is a deeper, jointly-built and jointly-marketed effort. They're all "partnerships," but each has different owners, economics and risk, so the first decision is which one you're actually building.

Who should own partnerships, sales, BD, or a separate team?

The failure mode is leaving it unowned, or bolting it onto a salesperson who is also carrying a direct quota and will always favour the deal they control. Lightweight referral programs can sit close to marketing; serious co-sell and alliance motions usually need a dedicated partner owner with their own number and a clear interlock with direct sales to manage channel conflict. Decide explicitly, and measure the owner on partner-sourced and partner-influenced revenue.

How do I avoid channel conflict with my own salesforce?

Define rules of engagement before you launch, not after the first contested deal. Decide which segments, products or geographies are partner-led versus direct, how a deal gets registered to a partner, and how commissions are split when both touch an account. The goal is for a direct rep to see a partner as help, not competition, which usually means the comp plan rewards them when a partner closes business in their territory rather than penalising them.

Why do so many partnerships produce no revenue?

Almost always because the partnership was treated as an announcement rather than a channel. There was a signed contract and a launch, but no owner, no pipeline, no joint account planning, and, most commonly, no real reason for the partner to prioritise you over their other options. Ernst and Bamford's near-50% success rate is largely a story of relationships that were never resourced and managed like the revenue channel they were meant to be.

How do I know whether a partnership is actually working?

Track partner-sourced pipeline (deals the partner originated), partner-influenced revenue (deals they helped close), and the health of the shared customer (do referred customers activate and retain better?). Pair those with a simple two-sided check: is the partner making money or serving their customer better because of you? A partnership that shows up in your pipeline but not in the partner's is one you're about to lose.

Related in the Toolkit

Where to go next