AVODA Group

Menu

Co-build the Floor, Compete on the Ceiling

By /

When it pays to partner with a competitor — and how to do it without handing them your future.

The question under the question

Most leaders frame the choice as a binary: compete or cooperate. That framing is the first mistake. You compete to divide value; you cooperate to create it. So the real question is never “are they a rival or a partner?” — it is “in this arena, right now, is most of the value still uncreated, or already on the table being split?”

Get that one diagnosis right and the rest follows. Partner when the advantages are complementary and the value is still uncreated. Compete when the advantages substitute and the value already exists. Everything in this essay is an elaboration of that single sentence — and a method for acting on it.

The stakes are not academic. Strategic alliances now sit at the centre of how value is built, yet the failure rate is brutal: most estimates put alliance termination between 60% and 80% (Peter Simoons; FasterCapital). Partnering with a competitor is the highest-variance version of this game — the upside is a bigger market, the downside is that you train the firm that later eats you. The difference between the two outcomes is rarely the idea to partner. It is where you partner.

The doctrine: a market has a floor and a ceiling

Here is the framework I want you to leave with. In any market the value chain splits into two layers.

The Floor is everything customers never choose you for: the shared, undifferentiated, cost-heavy, table-stakes layer — rails and standards, compliance plumbing, basic infrastructure, foundational R&D, and the work of educating the market that a category should exist at all. The Floor is expensive to build and identical no matter who builds it.

The Ceiling is everything customers do choose you for: the differentiated layer — the experience, the brand, the craft, the relationships, the proprietary insight. The Ceiling is where your margin and your moat live.

The doctrine is one line: co-build the Floor, compete on the Ceiling.

The two classic strategic errors are now obvious. Competing on the Floor means duplicating cost that no customer will ever reward you for — two firms laying parallel railway track to serve the same village. Cooperating on the Ceiling means leaking the very thing that makes you you. Winners do the opposite: they pool the undifferentiated cost and race each other on the differentiated value.

This is why bitter rivals are so often each other’s biggest customers and suppliers. Samsung has supplied the OLED display for every Apple phone that used the technology, from the iPhone X onward — and in 2017 Samsung reportedly earned more from Apple’s iPhone X than from its own flagship Galaxy S8, selling Apple on the order of 180–200 million panels (Android Authority; Substack analysis). Apple and Samsung compete ferociously on the Ceiling — the phone you choose — while Samsung quietly supplies a critical piece of the Floor. For Apple’s foldable, the two even signed a three-year exclusive supply deal (SamMobile). Neither firm confused the layers.

The same logic explains the airlines. Star Alliance — founded in 1997 — grew to roughly 23% of global scheduled traffic by 2015, spanning 1,200+ destinations and 16,000+ daily departures (Wikipedia: Airline alliance). Member airlines cooperate on the Floor (network reach, codeshares, lounges, baggage standards) and compete on the Ceiling (fare, service, schedule). The model worked so well that rivals copied the structure, spawning Oneworld (1999) and SkyTeam (2000). And in electronics, Sony and Samsung ran the S-LCD joint venture to share the brutal capital cost of a panel fab (Floor) while competing screen-to-screen in the living room (Ceiling).

Why the doctrine works: the established theory

The Floor–Ceiling doctrine is a practical compression of three respected bodies of work. It pays to know them, both for rigour and to borrow their tools.

The Value Net (Brandenburger & Nalebuff, 1996). Their great contribution was to add a fourth player to the board: alongside customers, suppliers, and competitors sits the complementor — anyone who makes your offering more valuable. Competitors shrink your added value; complementors grow it. The same company can be both at once, on different layers — which is exactly the Floor/Ceiling split. Their PARTS levers (Players, Added value, Rules, Tactics, Scope) are a checklist for changing the game rather than merely playing it (MindTools; HBS Online).

The Relational View (Dyer & Singh, 1998). Their insight — that competitive advantage can live between firms, not only inside them — identifies four sources of “relational rents”: relation-specific assets, knowledge-sharing routines, complementary resources and capabilities, and effective governance (Academy of Management Review). Two of these four are your go/no-go tests (is it complementary?) and your make-or-break execution variable (is it governed well?).

Transaction Cost Economics and the Resource-Based View. TCE (Williamson) reminds you that every partnership is also a governance problem — the gains can evaporate in coordination cost and opportunism. RBV (Barney) reminds you to never share the resource that is your advantage. The Relational View sits on top of both, extending the resource lens beyond your own walls.

The throughline: cooperate where resources are complementary and governable; compete where the resource is your moat. Floor and Ceiling.

The decision: four gates

A framework without a decision rule is a poster. So here is the instrument I use. A partnership with a competitor is wise only if it passes four gates, in order:

  1. The Pie Gate — is the value mostly uncreated? If the category is young, thin, or stalled, the real enemy is customer inertia, not the rival. Two firms building a market grow it faster than one. (If the pie is mature and fixed, stop — this is a share fight; compete.)
  1. The Complementarity Gate — are our advantages complementary, not substitutive? If you’d both bring the same thing to the same customers, there is nothing to trade and everything to leak. (If substitutive, compete.)
  1. The Net-Leak Gate — will we capture more than we leak? Every competitor partnership is a controlled transfer of capability. Ask honestly: in this deal, who is the student and who is the teacher? (If they learn your edge faster than you learn theirs, walk.)
  1. The Dependence Gate — does it keep dependence symmetric? Partner from strength, not need. If the structure leaves you more dependent on them than them on you, they can squeeze you later. (If asymmetric against you, stay arm’s length.)

Pass all four, and you cooperate — then choose the structure by trust and stakes (see below). Fail any one — especially the last — and you compete or keep your distance. Most disasters are firms that loved the upside of gate 1 and never ran gates 3 and 4.

Matching structure to stakes

Cooperation is a spectrum, not a switch. From loosest to tightest: a standards body → a data/interoperability pact → co-marketing → joint R&D → a joint venture → an equity stake → a merger. The error is over-binding a thin overlap (a full JV where a codeshare would do) or under-structuring a deep one. And because 60–80% of alliances die on governance — unclear decision rights, no executive sponsor, incompatible objectives, no exit clause — the boring structure decides the outcome more than the brilliant rationale (FasterCapital). Ring-fence the crown jewels with clean rooms and scoped data; write the renegotiation in on day one.

The East African proof: thin markets reward cooperation

Nowhere is the doctrine clearer than in African financial services, because the markets are young — which means the Floor barely exists yet, and whoever builds it together wins.

Start with the scale of the prize. McKinsey projects African fintech revenues could grow roughly eightfold, from about $4 billion in 2020 to around $30 billion by 2025, inside a financial-services market reaching about $230 billion (McKinsey: Fintech in Africa). Globally, mobile money crossed $1.4 trillion in annual transaction value with 1.75 billion registered accounts, and Sub-Saharan Africa is the engine (GSMA).

Now the coopetition. M-Pesa holds roughly 96.5% of Kenya’s mobile-money market and serves over 60 million customers a month (Market Data Forecast). A dominant player like that could have walled its garden forever. Instead, market-led interoperability took effect in Kenya in 2018, letting customers move money across providers — and in Kenya, Rwanda, and Tanzania, provider-led interoperability raised adoption and average transaction values rather than cannibalising them (GSMA). The rails are the Floor; everyone agreed to share them, and the pie grew.

Layer on the partnerships: MTN and JUMO co-built Qwikloan; Fingo partnered with Ecobank; Nigeria’s NearPays chose alliances with Fidelity Bank Ghana and MTN over going it alone; MTN’s MoMo and Ecobank co-built instant SME micro-loans that bypass collateral (Telecom Review Africa; UCT). And at the continental scale, PAPSS — the Pan-African Payment and Settlement System, built under the AfCFTA to connect 50+ countries and ~40 currencies — is the Floor as industrial policy, with the first Pan-African card scheme, PAPSSCARD, launched in June 2025 (McKinsey: Future of payments in Africa).

The lesson for any East African founder or institution: in a thin market, your competitor is rarely the threat — the absence of a market is. Co-build the rails, the standards, and the customer education; compete on the product that runs on top.

The AI age changes one variable — and it’s the decisive one

Here is the twist that makes this urgent in 2026. AI lowers the ceiling. Capabilities that were differentiators last year — drafting content, writing boilerplate code, first-line support, basic analysis — are being commoditised into table-stakes. What was Ceiling is collapsing into Floor.

You can watch the doctrine play out at the frontier. Microsoft put at least $13 billion into OpenAI (over 16% of its disclosed financing); Amazon put at least $8 billion into Anthropic (over 25%) — and AWS now funds both OpenAI and Anthropic, while Anthropic’s latest mega-round drew in roughly a dozen investors who also back OpenAI, Microsoft among them (CNBC; TechCrunch). Read it through the doctrine: the cloud giants cooperate on the Floor (compute, capital, data-centre capacity) while their model partners compete on the Ceiling (the frontier model). Even the consultancies show it — AI and tech now drive ~40%+ of revenue at both McKinsey and BCG, and both deliver it by partnering with Microsoft and Google Cloud rather than building hyperscale infrastructure themselves (McKinsey at 100).

Three AI-age rules fall out of this:

  1. Re-map your Floor and Ceiling every quarter, not every decade. AI keeps pushing the line up. The capability you guard today may be free next year.
  2. Cooperate on the newly-commoditised Floor — shared models, compute, data standards, common safety and evaluation practices — and race to a higher Ceiling: judgment, trust, proprietary data, relationships, taste, accountability. These are the things AI cannot copy from your competitors because they are not in the training data.
  3. Let AI run your Floor so your scarce human time compounds on your Ceiling. This is as true for a one-person business as for a hyperscaler.

Action plans: decisions and first moves, by tier

(Decisions to make, then the first concrete actions — framed for the age of AI.)

Big business

Decisions. Where will you set or co-own the standard before someone sets it for you? Which non-core capex (infrastructure, foundational R&D, compliance utilities, AI compute) should you pool with a rival rather than duplicate? Where do you take an equity stake in an adjacent competitor versus build? Where is the antitrust line?

Actions (first 90 days).

  • Run a Floor Map of your full value chain: tag every activity Floor (shared, undifferentiated, cost-heavy) or Ceiling (chosen, differentiated, defensible). Cooperate down the Floor list; invest up the Ceiling list.
  • Stand up an alliance-governance function — decision rights, an executive sponsor, an exit clause — because governance, not strategy, is what kills 60–80% of alliances.
  • For AI: cooperate on compute and foundation models; compete on proprietary data and workflow. Negotiate non-exclusive cloud and model terms so no single partner can throttle your reach.
  • Ring-fence crown jewels with clean rooms and scoped data before any competitor data exchange.

SMEs

Decisions. Which shared infrastructure do you join rather than build (payments, logistics, KYC, AI tooling)? Which 2–3 peers could you form a “category coalition” with to educate the market? Where do you partner to reach distribution or capital you cannot build alone?

Actions (first 30 days).

  • Join the rails, don’t rebuild them — plug into existing payment, mobile-money, and logistics networks instead of reinventing the Floor.
  • Form a category coalition with two or three respected peers: co-author a standard, co-host an event, co-publish the buyer’s guide that grows the whole category.
  • Ring-fence your client relationships — keep the Ceiling (your service, your trust) entirely your own even while you share the Floor.
  • Put AI on the Floor (ops, drafting, scheduling, first-line support) and reinvest the freed hours into the human Ceiling — relationships, judgment, follow-through.

Micro-business

Decisions. Which “competitor” next door is actually a complementor for shared fixed costs? When do you refer overflow instead of turning it away? What can AI do so you stop doing it by hand?

Actions (this week).

  • Form a buying or referral group with nearby peers — bulk-purchase inputs, share a delivery rider or a market stall, swap overflow customers for a referral fee. You shrink the Floor cost together and still compete for the walk-in.
  • Co-host — a joint promotion, a shared pop-up — to pull more customers into the category than either of you could alone.
  • Let AI handle the Floor: bookkeeping, quotes, captions, customer replies. Spend the time you save on the one thing the shop next door can’t copy — the relationship with your customer.

The one idea to keep

In young or thin markets — which is most of East Africa, and increasingly most of the AI economy — treat your competitor as a co-builder of the category and a rival only at the surface the customer sees. Co-build the floor; compete on the ceiling. Run the four gates before you sign. And remember coopetition has a half-life: as the market matures the relationship tilts back toward competition, so write the renegotiation in on day one.

Sources

Note on the numbers: the 60–80% alliance failure figure is a widely-cited estimate range, not a single audited statistic; market-size projections (McKinsey fintech, GSMA mobile money) are forecasts and should be cited as such.

Leave a Comment

Your email address will not be published. Required fields are marked *