Diversification and Advantage
Martin’s rule for corporate portfolios: you can't use a competitive advantage twice.
An advantage is specific to a Where to Play / How to Win pair. Carrying a brand, capability or asset into an adjacent business doesn't carry the advantage: the new arena has its own customers, rivals and economics, and demands its own cascade. Most diversification quietly assumes the advantage transfers. Most write-downs prove it did not, and the write-down usually arrives years after the assumption was made. That's why the lesson never seems to stick.
The portfolio discipline
Three rules. Each business must have its own defensible cascade, audited with the Opposite Test and the Can't-Won't Test, not a paragraph explaining how it "leverages the core". Sell businesses you don't love: an owner who doesn't love a business under-invests in its capabilities and systems, so the business is worth more to someone who does, and holding it's a slow tax on both parties. And corporate centres add value only where they genuinely strengthen a business's must-have capabilities; otherwise they are cost plus interference, the same earn-your-place logic applied to shared functions in Functional Strategy.
The apparent exceptions
Amazon and Google look like refutations, advantage apparently reused across arena after arena. Our compound research reconciles the exception: expansion that feeds the same compound engine, the same data loop, the same infrastructure flywheel, amplifies the advantage rather than reusing it, while expansion that assumes the advantage transfers without the engine is exactly the write-down this rule predicts. The engine test, and how to tell which kind of expansion you're contemplating, lives in Compound Advantages.
How this connects to OKRs
Diversification decisions produce a recognisable OKR pathology: the new venture's goals borrowed from the core business's playbook, measuring the advantage the venture was assumed to inherit rather than testing whether it exists. The honest version runs the new arena as a bet: Key Results that test the new cascade's pivotal conditions, what would have to be true for the advantage to hold here, answered by evidence rather than by the annual report's optimism.
One line to keep: Carrying an asset into a new arena doesn't carry the advantage; the new arena demands its own cascade.
Our synthesis of Roger Martin’s published work, sources credited. Read the originals: they’re excellent.
Why does diversification so often destroy value? +
Because advantage is specific to a where-to-play/how-to-win pair. Carrying a brand or capability into an adjacent arena doesn't carry the advantage; the new arena has its own customers, rivals and economics, and demands its own defensible cascade.
What about Amazon and Google? +
They feed the same compound engine, the same data loop or infrastructure flywheel, so expansion amplifies the advantage rather than reusing it. Expansion that assumes the advantage transfers without the engine is exactly the write-down the rule predicts.
Reading about method isn't the same as running it. We install this system and build the capability that stays.