Attack by Stratagem

Core idea

Attack by Stratagem contains the most famous proposition in the entire book: the supreme excellence is not to win a hundred battles — it is to subdue the enemy without fighting at all. Sun Tzu organises the topic around a hierarchy of attack, descending from the most desirable target to the least:

1. Attack the enemy's plans — defeat them at the strategic level, before contact.
2. Attack their alliances — isolate them politically.
3. Attack their army — engage in open battle.
4. Attack their fortified cities — the worst option, undertaken only when no alternative exists.

The topic then gives a brief catalogue of troop-ratio guidance (encircle at 10:1, attack at 5:1, divide at 2:1, evade if much inferior), and closes with the most-quoted single line in the book: "Know the enemy and know yourself, and in a hundred battles you will never be in peril."

Sun Tzu's argument: "Best in war is to target their strategies, then their alliances, and then their troops. The least desirable option is to attack a fortified city."

Why it matters

A theory of where to apply force

Most of strategy is the question where to apply force, given that force is finite. Attack by Stratagem gives a clear answer: as high up the chain of causation as possible. By the time you are at the city walls, you have failed three times — you did not break their plans, you did not break their alliances, you did not break their army in the field. Force applied at the strategic level multiplies; force applied at the tactical level grinds. This is the foundational insight behind every later doctrine of "centre of gravity" — Clausewitz, Boyd, modern competitive strategy.

The cost of winning poorly

Sun Tzu adds a subversive corollary: "Winning a hundred victories in a hundred battles is not the best possible outcome." Even uninterrupted winning can be a sign that you have organised your effort at the wrong level. A hundred won battles still means a hundred battles fought — and Waging War already established what each battle costs. The right metric is not victories per battle; it is victories per fight avoided.

Self-knowledge is half the calculation

The closing maxim has a three-state truth table:

• Know enemy and self: safe in a hundred battles.
• Know self, not enemy: win and lose by turns.
• Know neither: lose every battle.

The first half — know yourself — is the half most people skip. The enemy is exotic and interesting; one's own organisation is familiar and dull. Sun Tzu's structure insists the harder, less glamorous knowledge is just as important: an honest map of your own strengths, weaknesses, and constraints.

Key takeaways

Mental model

Copy sourceZoom

The hierarchy of attack

Attack their plans

The highest-leverage target is the opponent's decision-making process before it has finalised. If you can spoil their plan in its formative stage — by making it look unworkable, by removing an assumption it depends on, by getting them to commit to a different plan — you have won without firing a round. In modern terms: stealing the leak ahead of the opposing campaign, getting the rival's strategy committee to fixate on the wrong threat, making your own moves so unpredictable that no clean plan against you can be drawn.

Attack their alliances

The second-best target is their coalition — the relationships and dependencies that supply them with strength. An isolated opponent is a weaker opponent. Apply pressure to allies, suppliers, customers, regulators — the people around the rival who give them their advantage — and you weaken the rival without engaging them directly. Many corporate "wins" are actually alliance victories: the channel partner you signed exclusively, the regulator you turned, the talent you starved them of.

Attack their army

Engaging the army in the field is the acceptable but expensive option. It is the option Sun Tzu is willing to discuss in detail (the rest of the book is mostly about this), but he places it third in the hierarchy on purpose. If you find yourself here, you have failed to defeat them at the two higher levels.

Attack their cities — only when no alternative exists

The siege is the worst operation in the book. Sun Tzu is quantitative about it: three months to build siege engines, another three for the earthworks, and if the impatient commander sends troops up the walls "like ants," he loses one-third of his men and still fails to take the city. The lesson generalises: when an objective is so well-fortified that storming it consumes a third of your force, the objective is the wrong one. Find a different target.

Calibrating force to odds

Sun Tzu's force-ratio rules are blunt but useful:

| Numerical ratio | Recommended action | | --------------- | ------------------------------------------------ | | 10:1 | Encircle — surround completely | | 5:1 | Attack — direct engagement | | 2:1 | Divide — split the enemy and engage in parts | | 1:1 | Meet on the battlefield — even contest | | Inferior | Give ground — withdraw | | Much inferior | Evade — refuse engagement |

The most counterintuitive line is the last: "A smaller force's stubbornness in sticking to a plan makes it easy prey for the larger force." A small underdog who refuses to adapt — who insists on the original plan even when the asymmetry is brutal — is delivering themselves to the larger force.

The ruler's three errors

The ruler imperils his own armies in three specific ways:

1. Ordering advance or retreat without understanding the field — "hobbling the troops."
2. Governing the army with administrative rules suitable to a court — confusing the officers.
3. Failing to balance and synchronise operations — causing the officers to doubt his competence.

The implicit principle: the ruler's job is to appoint the right commander, not to play commander. "Inducing chaos in the army and throwing victory away" is what happens when the ruler keeps reaching past the commander into operational decisions. This is the ancient version of the executive who keeps redesigning the engineering org chart from above.

Practical application

Audit your effort by level of attack

For any contested initiative, list what you are doing and tag each item with one of the four levels. If most of your effort is at level 3 (engaging the opponent's army) or level 4 (storming their city), you are probably under-investing at levels 1 and 2. Move some effort up the ladder: how can you spoil their plan, weaken their alliances, before next quarter?

Make the self-audit honest

The know-yourself half of the famous maxim is best operationalised as a written, dated audit:

1. What are our actual current strengths (not branding)?
2. What are the constraints we will not admit out loud?
3. Where are we one bad month from a crisis?
4. Who on the team is over-extended, who is under-used?

The audit is most valuable when it surfaces something uncomfortable. If nothing uncomfortable appears, the audit was not honest.

Refuse the siege; reframe the objective

When you find yourself stuck on a fortified objective — a customer who won't move, a hire who won't sign, a regulator who won't budge — pause and ask: is this actually the right target, or is it the one I happen to be in front of? The Attack by Stratagem reading is that an objective requiring a months-long siege has usually been mis-identified. The real target is something else, easier to reach, that makes the original objective irrelevant.

Example

A SaaS startup is losing developer mindshare to an open-source competitor. The natural reaction is the siege: ship every feature the competitor ships, write blog posts comparing them point by point, run paid campaigns against their search terms. This is attack at level 3, almost level 4 (storming the walls of their feature set).

The Attack by Stratagem reading would step up the ladder. Attack their plans: the open-source competitor's roadmap depends on the contributions of a small core team — what if the team's attention is captured by something more interesting (an open-source project the startup sponsors that the same core team contributes to)? Attack their alliances: the competitor's strongest distribution is a single popular framework — what if the startup builds a deep integration with the framework's next major version before anyone else? Both moves leave the competitor's army intact but weaken the position from which they fight. The startup never engages the feature-for-feature siege at all.

Related lessons