Everything so far was about winning a slot in one mind. Part II is about choosing which battles to fight at all — and most marketing plans, on inspection, contain no strategy whatsoever.
Here's the whole chapter in one line: tactics is doing things right; strategy is choosing which things are worth doing — and a challenger only wins by narrowing the fight until it's locally the bigger army. Everything below is that sentence, with the receipts.
Open a typical marketing plan and here's what you find: a revenue goal, a budget, and a channel list with percentages next to it. Twelve slides of activity. It looks rigorous, it gets approved, and it contains no strategy at all — it's a to-do list wearing a strategy costume. Nothing in it says why these moves and not others, what problem they solve, or what the plan is willing to lose.
Richard Rumelt spent Good Strategy Bad Strategy (2011) making one distinction stick. A real strategy has a kernel of three parts, and in plain words they are: a diagnosis — what is actually going on here, the honest one-paragraph account of the situation; a guiding policy — the overall approach chosen to deal with it, which necessarily rules other approaches out; and coherent actions — a handful of moves that reinforce each other instead of competing for the same budget.
If you write software, you already know this document. A strategy kernel is an architecture decision record: the context (diagnosis), the decision (guiding policy), the consequences you accept (coherent actions and their costs). Nobody confuses a Jira backlog with an ADR. Marketing confuses them constantly.
And the cost of the confusion is precise: tactics without strategy is optimizing a function you never chose. The team runs gradient descent beautifully — better subject lines, cheaper clicks, tighter funnels — on a loss function nobody ever argued for. You converge, efficiently, somewhere that doesn't matter.
In 1986, Ries & Trout — the positioning duo from Chapter 1 — published Marketing Warfare, borrowing from the Prussian war theorist Clausewitz. Its lasting contribution is a map: your strategy is determined by your position in the market, not by your ambition. There are only four ways to fight, and which one is yours is mostly not a choice.
Defense is for the leader — and only the leader. The best defensive move is to attack yourself before anyone else can: Gillette famously shipped the razor that obsoleted its own best-seller, again and again, so every challenger arrived to find the leader already occupying the improvement.
Offense is for a strong #2. You don't attack the leader's weakness — you attack the weakness inherent in its strength, the flaw it cannot fix without dismantling what makes it the leader. Scope attacked Listerine's strength (the powerful medicinal formula) at its inherent cost: "medicine breath." Listerine couldn't answer without abandoning its own position.
Flanking is for the mid-sized player: move into uncontested ground and be first there. Miller Lite didn't brew a better Budweiser; it flanked into light beer, a territory the heavyweights had left empty, and led it for a decade.
Guerrilla is for the small: find a pond small enough to defend and be the biggest fish in it. In-N-Out has stayed a regional burger chain for seventy years, on purpose — inside its territory it out-loves, out-queues, and out-margins chains a hundred times its size. Rule one of guerrilla war: never, ever act like the leader.
Underneath all four postures sits one piece of arithmetic, older than marketing: the bigger army usually wins the open field. Not always — but so reliably that betting against it is a hobby, not a plan. Two and a half thousand years of generals and one century of ad budgets agree: God is on the side of the big battalions.
Which sounds like despair for everyone who isn't the leader, until you notice the loophole. The principle of force is local. What decides a battle isn't total army size — it's who has more force at the point of contact. A challenger can't be bigger everywhere. It can absolutely be bigger somewhere, by narrowing the front until its whole force concentrates on ground the leader defends with a detachment.
This is why the most common challenger tactic — "let's outbid them on brand search, that's where the buyers are" — fails on contact. You've chosen the exact square meter where the leader's entire budget is standing. Every auction reprices to what they can afford. The winning version is the opposite move: own one channel, one niche, one claim completely — 100% of your force against the 3% of theirs that bothers to show up.
Programmer's version: you don't beat a complexity disadvantage with enthusiasm. If the incumbent wins at scale n, you don't fight at scale n — you shrink n. Pick the battlefield where your constant factors dominate. The simulator below runs the math; try the pairings the books recommend, then try the ones they warn about.
Roger Martin — who ran strategy with A.G. Lafley at P&G — compressed strategy into a cascade of five questions, and the two in the middle do nearly all the work: where will we play, and how will we win there? Not "what will we do" — where, and how. Every real strategy is a matched answer to both; every bad one answers one and hand-waves the other.
Where to play should feel familiar: it's Chapter 1 wearing a general's uniform. The perceptual map you dragged a dot around in — axes of price, attitude, occasion, whatever the category actually sorts by — is a where-to-play tool. Empty territory on that map is an undefended front. "Where to play" is choosing which minds, which occasions, which channels, which price tier you're fighting for — and, just as loudly, which ones you're not.
How to win is the part plans skip, because it requires naming an actual advantage — something true of you that the competitor can't cheaply copy. There are only a few honest species of it: cost (you can profitably charge less), distinctiveness (Chapter 6's assets — you're recognized and recalled faster), distribution (you're where buyers are and rivals aren't), or product (a real, demonstrable difference the customer can feel). If your how-to-win names none of these, it isn't a how-to-win; it's a hope.
The two answers must agree. A premium where-to-play with a cost how-to-win is a contradiction wearing a brand book. And "everywhere, with excellence" fails both questions at once — it's the null strategy, syntactically valid and semantically empty.
Rumelt's other gift is a field guide to the impostors. Bad strategy isn't the absence of strategy — it's a set of recognizable species that occupy the slot where strategy should be, so nobody notices the vacancy. Four tells:
Fluff. Word salad that restates the domain in gassy abstractions: "leveraging synergies to deliver customer-centric value." The test: negate it. If no sane company would pursue the opposite ("customer-hostile value"), the sentence carries zero information.
Failure to face the problem. A strategy with no diagnosis. If the document never says what's hard — the competitor eating your share, the channel that died, the reason customers leave — then nothing in it can be a response to anything.
Goals mistaken for strategy. "Grow 30% year over year" is not a strategy; it's a wish with a spreadsheet. Ambition is the input to strategy, not the output. A target tells you the scoreboard; it doesn't tell anyone what to do differently on Monday.
Dog's-dinner objectives. A sprawl of twenty "strategic priorities" — which is to say, none. Choosing is the job. A list that includes everything is a list that decides nothing.
Marketing has its own dialect of each. "Be more premium, yet more accessible" — two directions at once, a refusal to choose dressed as nuance. "Target Gen Z and boomers" — everyone is not a segment. You now speak the dialect; the detector below checks your accent.
Fine words. Does any of this appear in measurable outcomes? Unusually for marketing — yes, and with numbers attached. The IPA databank holds hundreds of effectiveness-award case histories with audited business results, and Les Binet and Peter Field spent two decades mining it. Three findings matter here.
One clear strategy beats many. Campaigns built on a single-minded strategy — one diagnosis, one job — reliably out-produce campaigns juggling multiple objectives. The dog's-dinner tell isn't just ugly; it's measurably expensive.
The 60/40 rule. Marketing spend does two different jobs on two different clocks: brand building (slow, emotional, compounds for years) and activation (fast, targeted, harvests demand that already exists). The databank's best performers balance them at roughly 60% brand / 40% activation — flexing by category, more activation for direct businesses, more brand for subscriptions — and the failure mode is always the same direction: over-rotating to activation because it's the half that shows up in this quarter's dashboard.
ESOV — the cleanest strategic budget rule marketing has. Take your share of the category's advertising noise (share of voice) and subtract your share of its sales (share of market). That gap — excess share of voice — predicts growth: sustain a positive gap and share drifts toward your voice. It's the principle of force with a P&L: growth is bought with an overweight of force, and maintenance is priced at parity. Turn the dials below and watch the drift.
An obvious objection: all this talk of annual battle plans — hasn't the feed killed it? Media re-optimizes hourly, creative ships weekly, budgets move daily. Who needs a strategy when the machine adapts in real time?
Exactly backwards. The annual plan died; the kernel didn't — and the faster the tactics, the more the kernel matters. An always-on system with no standing guiding policy doesn't adapt; it thrashes. Each week's winning variant drags the brand a random step: this ad worked, so lean edgy; that one worked, so lean premium; after forty optimizations you've random-walked away from any position at all. Optimization without strategy is a hill-climber with no objective function — it will happily climb the nearest hill, which is rarely the one you needed.
The working modern rhythm splits the clocks. The kernel is standing — diagnosis, guiding policy, the sacrifice — revisited on a slow cadence, quarterly at most, and only changed for a reason you can write down. Tactics run at feed speed underneath it, and the kernel's real job is to act as a filter: which of this week's "winners" are actually allowed to win.
One more inversion worth naming: creative strategy is the new media strategy. Targeting converged — everyone buys the same auctions from the same three platforms, so where you show up is no longer an edge. What diverged is what you show. The choosing that used to happen in the media plan — which battles, which fronts — now mostly happens in the creative brief. That's where the war planning moved.
Everything in this chapter fits on one page, and if it doesn't fit on one page it isn't finished. Before the next planning cycle, write this — in sentences, not bullets of nouns:
That page is the ADR. Everything downstream — the channel mix, the creative brief, the promo calendar — is implementation, free to change weekly precisely because the page doesn't.
Which raises the implementation question: once you've chosen the war, what do you actually turn? It turns out you control exactly four knobs — and the ad, famously, is only part of one of them. That's Chapter 8.