Most "partnership strategies" are really just a procurement list with nicer logos. A true ecosystem is something different: a deliberate structure in which independent companies build on top of what you offer, and every one they add makes your product more valuable to the next customer, without you paying for it.

The quick version

  • An ecosystem is a set of companies that co-evolve around a shared offering; a platform is the product or rules that let them plug in. Partnerships are the deals that wire them together.
  • The prize is network effects: each new partner or user raises the value for everyone else, so growth compounds instead of just adding up.
  • The evidence is sobering. A BCG analysis of more than 80 business ecosystems found fewer than 15% are sustainable in the long run, most die from how they were designed, not how hard the team worked.
  • The leader's job isn't to sign the most deals. It's to decide what role you play (orchestrator or participant), make joining attractive enough that partners profit, and govern the system so it stays healthy.

The idea in depth

The vocabulary here is loose in everyday use, so it's worth being precise. The word ecosystem entered business strategy with James F. Moore's "Predators and Prey: A New Ecology of Competition" (Harvard Business Review, May–June 1993). Moore's argument was that a company should be seen not as a member of a single industry but as part of a business ecosystem that cuts across many, a community of organisations that co-evolve their capabilities around a new innovation, competing and cooperating at the same time. Think of how chipmakers, app developers, accessory brands and carriers all rose and fell together around the smartphone.

A platform is the narrower, more designed thing at the centre. In "Pipelines, Platforms, and the New Rules of Strategy" (HBR, April 2016), Marshall Van Alstyne, Geoffrey Parker and Sangeet Paul Choudary draw the sharpest line. A traditional "pipeline" business creates value by controlling a linear chain, source, build, sell. A platform creates value by orchestrating interactions between producers and consumers it does not own. Its critical asset is external: the community of participants. So the strategic job changes from controlling resources to coordinating them, and the firm wins by enabling more interactions and igniting network effects, the property that each extra participant makes the whole more valuable to the others.

So the move is: before you write a single partnership, decide which game you are actually playing. If your advantage comes from owning and optimising a chain, partnerships are supplements, useful, but not the engine. If your advantage could come from other people building on you, you are designing a platform, and your first question isn't "who do we sign?" but "what would make a third party choose to invest their own time and money on top of us?"

Pick your role: orchestrator, participant, or neither

Not everyone in an ecosystem plays the same part, and confusing them is expensive. Marco Iansiti and Roy Levien, in "Strategy as Ecology" (HBR, March 2004), name three roles. A keystone (what we'd now call the orchestrator) creates a platform others depend on and, crucially, shares the value it creates, keeping the whole system healthy. A dominator tries to own and absorb everything, which starves the partners it needs and eventually hollows out the very ecosystem it sits on. A niche player specialises and thrives by building on someone else's platform rather than running its own.

The keystone's discipline is generosity with a purpose: leave enough value on the table that partners keep showing up.

This matters because the most common strategic error is acting like a dominator while calling yourself a keystone, taking the orchestrator's central position but capturing so much of the value that complementors can't make money. App stores that quietly clone their most successful third-party apps, or marketplaces that squeeze sellers until the good ones leave, are dominators in keystone's clothing. The system looks fine on a quarterly dashboard and rots underneath.

The practical test: name your intended role out loud, then pressure-test it with one question, can a partner who does business with us build a healthy business of their own? If the honest answer is no, you are a dominator, and the ecosystem will eventually behave accordingly. And if you can't be the orchestrator, being a disciplined niche player on someone else's platform is often the better return than a doomed bid for the centre. Hagiu and Altman's "Finding the Platform in Your Product" (HBR, July–August 2017) is useful here: it lays out four concrete ways an ordinary product can open up to third parties, connect its customers, or plug into a bigger platform as a supplier, a reminder that "platform" is a set of design choices, not a birthright.

Why most ecosystems fail, and the honest limitation

Here is the part the strategy decks skip. Ecosystems are fashionable, and fashion hides a brutal base rate. Analysing more than 80 business ecosystems, the BCG Henderson Institute team of Ulrich Pidun, Martin Reeves and Maximilian Schüssler concluded that fewer than 15% are sustainable in the long run. Their related work argues most of the damage is done at the design stage, wrong governance, wrong incentives, the wrong problem, not in day-to-day execution. You cannot out-hustle a badly designed ecosystem.

The deeper limitation is one Cusumano, Gawer and Yoffie make plain in The Business of Platforms (HarperBusiness, 2019): network effects are real but they are not magic, and they are frequently overestimated. Many "platforms" never reach the critical mass where the flywheel spins on its own, and even strong ones can be undone by weak governance, multi-homing (partners and users happily using rivals too), or the slow erosion of trust. Treat ecosystem strategy as a high-variance bet, not a guaranteed moat.

What to do about it: before committing, do the unglamorous "why would this fail?" pass first. Be specific about the chicken-and-egg problem, which side do you attract first, and with what, before the other side exists to make it worthwhile? If you can't answer that in a sentence, you have a slogan, not a strategy.

flowchart LR
  A(["Pipeline business: you own the chain"]) --> B(["Add partners as suppliers / resellers"])
  B --> C{"Could others build value ON you?"}
  C -->|"No, you optimise the chain"| D(["Stay a pipeline. Partnerships = supplements"])
  C -->|"Yes, others can plug in"| E(["Design a platform"])
  E --> F(["Choose role: orchestrator (keystone)"])
  E --> G(["Or participate as a niche player"])
  F --> H(["Govern + share value → network effects"])
From pipeline to platform: the fork most teams skip. Leaders Loop

A worked example

Picture "Rota", a mid-sized maker of shift-scheduling software for cafés and restaurants. (The company and figures below are illustrative, to show the mechanics.) Rota has 4,000 paying venues and a flat growth curve. The founders want an "ecosystem play" and arrive with a list: integrate with three payroll providers, two point-of-sale vendors, and a delivery aggregator.

That list is the pipeline instinct in disguise, useful integrations, but each is a one-off deal Rota has to chase, build and maintain. The platform question reframes it: could outside developers build things on Rota that Rota would never build itself? Café owners keep asking for tip-splitting tools, allergen menus, local compliance checks. Rota can't build them all, and shouldn't.

So Rota chooses the orchestrator role deliberately. It publishes an open API and a simple revenue share, say developers keep 80% of what they charge venues, an illustrative figure chosen to keep building on Rota genuinely worth a developer's time (the keystone's "leave value on the table" discipline). It seeds the cold start by building the first two add-ons itself, so the marketplace isn't empty on day one, the answer to the chicken-and-egg problem. Within a year, suppose 30 third-party tools exist. Now every new café that joins finds a richer toolset (good for venues), and every new tool finds a bigger market (good for developers): a modest network effect, compounding. Rota's job shifts from shipping features to governing a marketplace, quality review, fair dispute handling, resisting the temptation to clone its best partners into dominator behaviour. The growth curve bends not because Rota worked harder, but because it changed what kind of company it was.

The counterfactual is the cautionary tale: had Rota launched the API with a 50% cut, vague rules and no seed apps, it would have an empty marketplace, no developers, and a board slide reading "ecosystem: in progress" for three years, a textbook design-stage failure.

Frequently asked questions

What's the difference between a partnership and an ecosystem?

A partnership is a bilateral deal, you and them, with defined terms. An ecosystem is a multilateral structure where many independent players build around a shared offering and, ideally, create value for each other, not just for you. Partnerships are the wiring; the ecosystem is the thing that lights up. You can have many partnerships and no ecosystem, which is the most common situation.

Do we need to be a tech company to have a platform strategy?

No. The pattern predates software, shopping malls, credit-card networks and franchise systems are all platforms that match two sides and grow on network effects. Hagiu and Altman's HBR work is largely about helping ordinary product companies find the latent platform in what they already sell. The technology lowers the cost of coordination; it doesn't create the logic.

How do we get past the chicken-and-egg problem?

Pick the harder side to attract and subsidise it, or fake one side until the other arrives. Common tactics: seed the supply yourself (build the first apps, recruit the first sellers), give one side a standalone reason to show up before the network exists, or focus on a single narrow niche where critical mass is small enough to reach. The one thing that never works is launching empty and hoping.

When should we NOT pursue an ecosystem?

When your real advantage is owning and optimising a chain, when you can't or won't share enough value to make partners money, or when you lack the patience and governance muscle for a multi-year, high-variance bet. Given that fewer than 15% prove durable, "stay a focused pipeline business" is frequently the wiser, more honest choice, and being a strong niche player on someone else's platform can beat a vanity bid for the centre.

How is this different from a joint venture or alliance?

Joint ventures and alliances are usually deep, exclusive arrangements between a few named partners with shared governance. An ecosystem is wider and shallower: many participants, lighter individual ties, open-ish rules of entry. They're complementary, see Joint ventures, alliances & strategic partnerships for when a tighter structure beats an open one.

Related in the Toolkit

flowchart TD
  K(["Orchestrator / keystone"]) -->|"shares value, sets fair rules"| P1(["Partner / complementor"])
  K -->|"shares value, sets fair rules"| P2(["Partner / complementor"])
  K -->|"shares value, sets fair rules"| P3(["Niche player"])
  P1 -->|"adds capability"| U(["Customers"])
  P2 -->|"adds capability"| U
  P3 -->|"adds capability"| U
  U -->|"more demand attracts more partners"| K
The network-effect loop a healthy keystone protects: value shared out comes back as demand. Leaders Loop

Where to go next

  • Pipelines, Platforms, and the New Rules of Strategy (Van Alstyne, Parker & Choudary, HBR, 2016), the clearest short read on why platform strategy is a different game from pipeline strategy.
  • The Business of Platforms (Cusumano, Gawer & Yoffie, HarperBusiness, 2019), the book for the full picture, including the honest limits and failure modes of platforms.
  • Strategy as Ecology (Iansiti & Levien, HBR, 2004), the keystone / dominator / niche framing that should decide which role you play before you sign anything.
  • Finding the Platform in Your Product (Hagiu & Altman, HBR, 2017), four concrete ways an ordinary product can open up into a platform, for when you suspect the latent one in what you already sell.