Alliance
Альянс
Collaboration Structure Depth → 2 · Excavator
partnership with independence kept
Moats it grows into
- → #11 Collaboration
Reached through Integrations: the alliance gets to Collaboration only after the partners have co-built a connection both products depend on and their teams run it as one workflow. Before that it is a contract with a notice period.
Summary
Two or more independent firms agree to work together on something specific — a joint product, a shared channel, a co-sold bundle — while each keeps its own roadmap and its own exit. It is a structure for sharing an edge, not for keeping one. A moat appears only if the joint work grows into something neither side can run alone.
How it works
An alliance trades a slice of independence for a capability the partner has and you lack: their distribution, their certification, their installed base, their engineers. Because each side keeps its own ownership, the deal costs little to sign and little to break. That is the point — and the weakness. The gain arrives on day one; the lock does not.
An edge appears where the partners' work becomes interlocked in practice: a co-developed integration that only makes sense with both products, a shared support desk, joint customers who bought the pair. Then leaving the alliance means rebuilding the interlock, and each partner's success pulls the other along. Without that interlock, an alliance is a press release with a logo wall. A rival can sign the same partner tomorrow; most partners multi-home, and the exclusivity clause that would prevent it is exactly what an independent party refuses to give.
Why it holds — or only takes
An alliance is how participants are arranged, not what keeps a rival out; the partner is free, the customer is free, the contract has a notice period. It does not hold. It reaches a moat through Collaboration alone: when the joint work becomes a workflow both firms' teams live in — shared roadmaps, co-owned integrations, joint accounts nobody wants to split — the arrangement turns into a team product neither can exit cheaply. That is depth 2 and under attack from AI, which compresses the teams the moat depends on. Everything else an alliance delivers — reach, credibility, a logo — is rented for the contract's term.
Do you have it
- You already have this if a partner's engineers have commit rights on your integration, and yours on theirs.
- Joint customers file one support ticket for both products and expect one answer.
- Your partner's release notes mention you more often than your own do.
- Ending the deal would cost a re-engineering, not a phone call.
How to dig it
- Pick a partner whose customers already ask for your product by name, and write the integration together.
- Put a shared artefact in the deal — a co-owned repository, a joint roadmap, a common account manager.
- Ask for co-selling before co-marketing; a partner that sells you is harder to replace than one that links to you.
- Renew on outcomes: joint customers won, not events attended.
How it erodes
- The partner signs your rival too; most partners multi-home and exclusivity is what they will not sell.
- An acquisition changes the partner's priorities, and the integration goes unmaintained.
- A platform absorbs the joint feature natively and the alliance's reason to exist disappears.
- Agents build the integration a partner would have brought, so the capability no longer needs a partner.
Solo, without capital
A solo builder can sign alliances this quarter: an integration partnership with a tool your users already run, a bundle with a peer, a listing in someone's marketplace. Each brings reach you could not buy. None of it holds: the partner does not depend on you and can list five rivals beside you. The Collaboration moat behind it needs a team-shaped product and a partner whose team lives in it, which a single person rarely builds in a quarter.
In nature, in business
- Nature
- —
- Business
- partnerships
Cases
- Star Alliance — Airlines that stayed separate companies pooled code-sharing, lounges and frequent-flyer miles so a passenger booked on one could fly the whole group.
- Nike and Apple — Nike and Apple co-developed the Nike+ sensor and later a Nike edition of the Apple Watch, each keeping its own brand and its own products.
- Microsoft and Intel — Microsoft and Intel never merged, but Windows and x86 were developed to each other's roadmaps for two decades and the pair set the PC's default.
Combines with
- Bundling — A partner's product is the natural thing to bundle yours with, and the bundle is what customers see of the alliance.
- Completeness — Alliances fill the gaps an all-in-one offer would otherwise have to build itself.
- Union — An alliance that outlives its trial period tends to formalise; a union is where a working alliance goes.
Fights with
- Secrecy — An alliance runs on shared roadmaps and shared code; a strategy built on protected knowledge cannot give a partner what the alliance needs.
- Highest bidder — Exclusivity bought at any price and a partnership that keeps independence want opposite things from the same partner.